MacTech 1 to 12/MacTech-vol-1-12.toast/Reference/the cmsp digests ('94-'97)/csmp digest Vol 3 No 135

Text File | 1996-02-09 | 144KB | 3,799 lines

C.S.M.P. Digest             Thu, 08 Feb 96       Volume 3 : Issue 135

Today's Topics:

        Command-Shift-Keys
        CustomPutFile Puzzle?
        Detecting QuickTime PowerPlug
        How can you use 68K MPW .o file in PPC apps?
        How do I create many menus from one MENU resource?
        MacApp newbie questions
        Matching struct definitions to TMPL resources
        Menu Manager limitations?
        Programming a PinnBall
        Q: When to call ReleaseResource?
        Quickdraw 3D intro??
        Starting MacTCP on startup
        VBL Afterlife
        Why would dragging in a thumb crash?
        [ANNOUNCE] BSP Tree Demo Application v1.3
        [Q] PowerPC VBLs?
        _Jackson Trap
        detecting CD-ROM



The Comp.Sys.Mac.Programmer Digest is moderated by Francois Pottier
([e-mail]).

The digest is a collection of article threads from the internet
newsgroups comp.sys.mac.programmer.help, csmp.tools, csmp.misc and
csmp.games. It is designed for people who read news semi-regularly and
want an archive of the discussions.  If you don't know what a
newsgroup is, you probably don't have access to it. Ask your systems
administrator(s) for details. If you don't have access to news, you
may still be able to post messages to the group by using a mail server
like anon.penet.fi (mail [e-mail] for more information).

Each issue of the digest contains one or more sets of articles (called
threads), with each set corresponding to a 'discussion' of a particular
subject.  The articles are not edited; all articles included in this digest
are in their original posted form (as received by our news server at
nef.ens.fr).  Article threads are not added to the digest until the last
article added to the thread is at least two weeks old (this is to ensure that
the thread is dead before adding it to the digest).  Article threads that
consist of only one message are generally not included in the digest.

The digest is officially distributed by two means, by email and ftp.

If you want to receive the digest by mail, send email to [e-mail]
with no subject and one of the following commands as body:
    help                                Sends you a summary of commands
    subscribe csmp-digest Your Name     Adds you to the mailing list
    signoff csmp-digest                 Removes you from the list
Once you have subscribed, you will automatically receive each new
issue as it is created.

The official ftp info is ftp://ftp.dartmouth.edu/pub/csmp-digest.
Questions related to the ftp site should be directed to
[e-mail].

-------------------------------------------------------

>From [e-mail] (Tim DeBenedictis)
Subject: Command-Shift-Keys
Date: Sat, 13 Jan 1996 21:24:27 GMT
Organization: NETCOM On-line Communication Services (408 261-4700 guest)

This is actually in response to me earlier "Why won't my command-keys
work?" question; many thanks to all of you who answered.  I now
understand the problem, and I have an idea for a solution.  Essentially,
here's the idea:

1) When the "command" key is down, and I get a key-down event, I can
manually check to see if the shift key is down by using GetKeys.

2) If the shift key is not down, I just pass the character code from the
KeyDown event to MenuKey, and everything works the way it does currently,
no problem.

3) If the shift key -is- down, I need to convert the key code ffrom the
KeyDown event to a character code, given that the shift key is now also
down.  The only missing piece is that I need to know how to do this.  I
am assuming that there's some built-in toolbox trap which does this, or
some other relaible method.

Any ideas?

Thanks again,

-Tim DeBenedictis


+++++++++++++++++++++++++++

>From [e-mail] (Dave Nebinger)
Date: Mon, 15 Jan 1996 14:49:28 -0500
Organization: KHP Services, Inc

In article <[e-mail]>, [e-mail] (Tim
DeBenedictis) wrote:

> This is actually in response to me earlier "Why won't my command-keys
> work?" question; many thanks to all of you who answered.  I now
> understand the problem, and I have an idea for a solution.  Essentially,
> here's the idea:
>
> 1) When the "command" key is down, and I get a key-down event, I can
> manually check to see if the shift key is down by using GetKeys.

And would this really work?  If the delay is too long from the point when
the key command is issued and when it is received, the shift key may not
be down.

Start a processor intensive task in the background and initiate an event
in the foreground.  By the time the app gets the event you could be off filling
your coffee cup.

Simply put, when your app receives and processes the event is completely
independent from the state of the keyboard.  That's the nature of event
processing and the multiprocessing system.

Dave.

==========================================================
Dave Nebinger                             [e-mail]
             The Alt.Sources.Mac Archivist
     <http://www.AmbrosiaSW.com/alt.sources.mac/>
    <ftp://ftp.AmbrosiaSW.com/pub/alt.sources.mac/>

+++++++++++++++++++++++++++

>From [e-mail] (Peter Castine)
Date: Tue, 16 Jan 1996 14:07:55 +0100
Organization: Process Control Center, TU Berlin

In article <[e-mail]>, [e-mail] (Tim
DeBenedictis) wrote:

> This is actually in response to me earlier "Why won't my command-keys
> work?" question; many thanks to all of you who answered.  I now
> understand the problem, and I have an idea for a solution.  Essentially,
> here's the idea:
>
> 1) When the "command" key is down, and I get a key-down event, I can
> manually check to see if the shift key is down by using GetKeys.

A better bet is to take a look at the modifiers flags in the Event Record.
A finite amount of time has passed between when the Command-Key event was
posted and when you call _GetKeys; things may have changed in that time.

I may have misdirected Tim in a former post I sent. The actual culprit in
the Command-Shift business is the _KeyTrans trap, which translates virtual
key codes into ASCII and ignores the Shift flag when the Command flag is
set. The modifier flags in the Event Record ought to correctly represent
what the user pressed.

> 3) If the shift key -is- down, I need to convert the key code ffrom the
> KeyDown event to a character code, given that the shift key is now also
> down.  The only missing piece is that I need to know how to do this.

_KeyTrans is the trap to use.

This was nicely documented in the old TechNote ``International
Cancelling,'' including more than sufficient source code. The information
has been folded into the new _Inside Macintosh: Text_, Appendix C (pp.
C-23 ff.).

Hope this helps,

Peter

- -------------- http://www.prz.tu-berlin.de/~pcastine/Music --------------
Dr. Peter Castine           | Grosze Potentaten und Regenten haben die
[e-mail]   | Pflicht, sich um die freien Kuenste zu
Process Control Center      | kuemmern. Zwar haben Privatleute Lust an der
Technical University Berlin | Musik und lieben sie, aber koennen sie nicht
                            | finanzieren.                -- Martin Luther

+++++++++++++++++++++++++++

>From [e-mail] (Bo Lindbergh)
Date: 18 Jan 1996 02:35:31 GMT
Organization: Royal Institute of Technology, Stockholm, Sweden

In article <[e-mail]> [e-mail] (Peter Castine) writes:
)
) I may have misdirected Tim in a former post I sent. The actual culprit in
) the Command-Shift business is the _KeyTrans trap, which translates virtual
) key codes into ASCII and ignores the Shift flag when the Command flag is
) set.

You're still wrong; KeyTranslate only follows orders.  The actual culprit
is the KCHR resource.


/Bo Lindbergh

+++++++++++++++++++++++++++

>From Brown Edward W -BSc Mod Comp <cme92024>
Date: 16 Jan 96 14:58:21 GMT
Organization: (none)

You can tell if the shift key is pressed from the modifiers field of the event
record. I believe you have to decode the key into a menu option yourself.

Ed.


+++++++++++++++++++++++++++

>From [e-mail] (Milo Shiff)
Date: 20 Jan 1996 04:25:29 GMT
Organization: Los Angeles Macintosh Group BBS

In article <[e-mail]>, [e-mail] (Tim
DeBenedictis) wrote:

> This is actually in response to me earlier "Why won't my command-keys
> work?" question; many thanks to all of you who answered.  I now
> understand the problem, and I have an idea for a solution.  Essentially,
> here's the idea:
>
> 1) When the "command" key is down, and I get a key-down event, I can
> manually check to see if the shift key is down by using GetKeys.

   I use shift and option keys for special modifiers to the command key
combinations in several of my programs.

   At offset $E (decimal 18) of the event record you will find the modifier
flags. Bit 11 is the option key, bit 9 is the shift key, and bit 8 is the
command key.

   Simply run a test on what combinations of modifier keys are being held
down and then dispatch accordingly.

   By the way, I do NOT recommend using _MenuKey to evaluate command keys.
You gain a lot of flexibility by keeping your own table. In addition to being
able to maintain command key combinations that don't appear in your menus,
you can more easily implement a user defined mapping of command key
functions.

- via BulkRate 2.1

---------------------------

>From [e-mail] (Roger Brown)
Subject: CustomPutFile Puzzle?
Date: 20 Jan 1996 16:30:25 GMT
Organization: Moonlight Software

I want to make what appeared to be a simple extension to CustomPutFile
but I can't figure out how to do it. When the user clicks the "Save"
button and the file name matches an existing file, I want to add an
"Append" option to the "Cancel" and "Replace" buttons that appear in
the standard "Replace" dialog.

Here are two things that I tried and the problems that resulted:

1. I tried a custom dialog hook and a custom dialog to use instead of
the "Replace" dialog. This worked fine by adding a different "Replace"
dialog with the same id as the System one in my program. My version has
an Append button and the dialog hook proc detects hits in that button.

The problem is that all other uses of SFPutFile variants in the program
use the "append-capable" replace dialog in my program when appending
makes no sense in their contexts. Is there a clean way to tell or trick
CustomPutFile to use my replace dialog only in one case? I don't want
to have to rewrite the resource map to fiddle with resource ID's every
time this procedure is accessed.

2. A much more involved approach is to use a Modal Dialog filter that
detects clicks on the Open/Save button and then does all the business
of seeing if the file name is already used. I think I can get the
information I want from the sfReply record if I access it through the
myDataPtr parameter. The problem here is that, contrary to what Inside
Mac says, the modal dialog filter gets events AFTER they are already
processed and does not get notification of the Open/Save button being
pushed at all!

Does anyone have any better ideas? Or tell me what I am missing here?

Thanks

- ------------------------------------------------------------------
Roger Brown                                [e-mail]
Moonlight Software
West Lebanon, NH, USA

+++++++++++++++++++++++++++

>From "Andrew C. Plotkin" <[e-mail]>
Date: Sat, 20 Jan 1996 14:02:41 -0500
Organization: Carnegie Mellon, Pittsburgh, PA

[e-mail] (Roger Brown) writes:
> I want to make what appeared to be a simple extension to CustomPutFile
> but I can't figure out how to do it. When the user clicks the "Save"
> button and the file name matches an existing file, I want to add an
> "Append" option to the "Cancel" and "Replace" buttons that appear in
> the standard "Replace" dialog.
>
> Here are two things that I tried and the problems that resulted:
>
> 1. I tried a custom dialog hook and a custom dialog to use instead of
> the "Replace" dialog. This worked fine by adding a different "Replace"
> dialog with the same id as the System one in my program. My version has
> an Append button and the dialog hook proc detects hits in that button.
>
> The problem is that all other uses of SFPutFile variants in the program
> use the "append-capable" replace dialog in my program when appending
> makes no sense in their contexts. Is there a clean way to tell or trick
> CustomPutFile to use my replace dialog only in one case? I don't want
> to have to rewrite the resource map to fiddle with resource ID's every
> time this procedure is accessed.

Rewrite the resource map? Not sure what you mean.

CustomPutFile has a dialogid argument. That's what you use to specify
a dialog resource. If you put your dialog resource (with the Append
button) in as resource 2000, and a normal dialog in as 2001, you can
switch back and forth just by supplying 2000 or 2001 in the dialogid
argument.

Is that what you meant?

--Z

"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."

+++++++++++++++++++++++++++

>From [e-mail] (Stefan Kurth)
Date: Sun, 21 Jan 1996 00:34:14 +0100
Organization: none

Roger Brown <[e-mail]> wrote:

> I want to make what appeared to be a simple extension to CustomPutFile
> but I can't figure out how to do it. When the user clicks the "Save"
> button and the file name matches an existing file, I want to add an
> "Append" option to the "Cancel" and "Replace" buttons that appear in
> the standard "Replace" dialog.

Write a dialogHook function.  Upon receiving sfItemOpenButton, check if
the file already exists.  If it does, put up your custom
Replace/Append/Cancel dialog.  If the user clicks Cancel, return
sfHookNullEvent from the dialogHook.  If the user clicks Replace or
Append, note this in a global variable, and return sfHookCancelButton.
This is important, because otherwise the system would put up its own
Replace dialog after yours.  So you'll need a global variable to
distinguish whether the user clicked cancel in the main dialog, or
Replace or Append in your custom dialog.


--
Stefan Kurth
Berlin, Germany

---------------------------

>From [e-mail] (Mark L. Hessenflow)
Subject: Detecting QuickTime PowerPlug
Date: 24 Jan 1996 12:08:36 -0800
Organization: Amstadt Consulting Group, San Jose, CA


I have some code which creates a preview picture for a file using the
QuickTime routines MakePreviewPicture() and AddFilePreview(). I check
with gestalt to see if both QuickTime and the image compression tools
are available before making the calls.

When running on the PowerPC, if QuickTime is installed but 'QuickTime
PowerPlug' is not, then the call to AddFilePreview() crashes.

In any other case, the code produces the preview if QuickTime is
installed, or doesn't if QuickTime is not installed.

Here is the code I use:

err = Gestalt( gestaltQuickTime, &attributes );
if( err == noErr ) {
        err = Gestalt( gestaltCompressionMgr, &attributes );
        if( err == noErr ) {
                picHandle = MakePreviewPicture( iconOffscreen );
                if( picHandle ) {
                        (void) AddFilePreview( resFile, 'PICT', picHandle );
                        }
                }
        }

So is there any way to detect 'QuickTime PowerPlug'?  Should I be doing
anything to make sure the shared library is loaded into memory?

-Mark

--
Mark L. Hessenflow
[e-mail]


+++++++++++++++++++++++++++

>From [e-mail] (Sanford H. Selznick)
Date: 24 Jan 1996 22:45:31 -0700
Organization: University of Arizona CS Department, Tucson AZ

Oh... this one...

bool QuickTimeIsInstalled(void)
{
  long            qtVersion;
  OSErr           anErr;
#ifdef powerc
  long            qtFeatures;
#endif

  anErr = Gestalt(gestaltQuickTime, &qtVersion);
  if (anErr != noErr)
    return FALSE;           // no QT present

#ifdef powerc
  // Test if the library is registered.
  anErr = Gestalt(gestaltQuickTimeFeatures, &qtFeatures);
  if ( !( (anErr == noErr)  &&
          (qtFeatures & (1 << gestaltPPCQuickTimeLibPresent))
        ) )  // not TRUE
    return FALSE;
#endif

  return TRUE;
}

I think I got this (or something similar) off a QuickTime Web page
over at Apple somewhere.

Hope this helps,
  Sanford

+++++++++++++++++++++++++++

>From [e-mail] (Mike Cohen)
Date: Thu, 25 Jan 1996 12:10:11 -0600
Organization: ISIS International

In article <4e63k4$[e-mail]>,
[e-mail] (Mark L. Hessenflow) wrote:

>When running on the PowerPC, if QuickTime is installed but 'QuickTime
>PowerPlug' is not, then the call to AddFilePreview() crashes.

I assume you have QuickTimeLib weak linked. If so, check the address of any
functions in it before you call them. You should do something like:

        if (AddFilePreview != NULL)
                AddFilePreview(...);
        else
                error("QuickTime PowerPlug isn't installed");

If you don't use weak link, your application will refuse to launch
completely if it isn't installed.


--
Mike Cohen - [e-mail]
Home Page: http://www.isis-intl.com/
Sound is the same for all the world - Youssou N'dour, "Eyes Open"
*** Save Mystery Science Theater 3000! Comedy Central has decided to
*** cancel it. Call them at [phone] and tell them what you think.
*** Full details at http://fermi.clas.virginia.edu/~jcp9j/canceled.html

---------------------------

>From [e-mail] (Jon Hewitt)
Subject: How can you use 68K MPW .o file in PPC apps?
Date: Thu, 18 Jan 1996 09:53:19 -0500
Organization: Liberty Mutual Group

Hi,

I have an API that requires a 68K MPW .o file which has no PPC library
counterpart. : * (

I assume that in order to be able to built a PPC app I will have to use
the Mixed Mode Manager.  But other than that, I'm not sure how to start.
Principally my problem is this:

How do you include the .o file in the application?  It doesn't seem like
either Metrowerks or Symantec (I would like to use both) allow you to
include .o files in PPC projects.  Do I need to build a code resource?  If
so, how do I call the functions in the API?

Does anyone know of sample code I could look at?


Thanks for any help you can offer,

Jon Hewitt

--
[e-mail]
My opinions are not necessarily those of my employer.

+++++++++++++++++++++++++++

>From [e-mail] (Etay Bogner)
Date: Sun, 21 Jan 1996 10:24:52 +0200
Organization: Scitex Corp.

In article <[e-mail]>,
[e-mail] (Jon Hewitt) wrote:

>> Hi,
>>
>> I have an API that requires a 68K MPW .o file which has no PPC library
>> counterpart. : * (
>>

It somewhat depends on how the functions in the .o are compiled and how
many of them do you have, but basically build a code resource.

You have several options on how to get to the functions inside the .o :

1) build a mixed mode entry per function and return their addresses via a
pointer to array of pointers to functions returned from the main of the
CR.
The client can define it's own functions that will jump to those pointers
or do a PPC asm hack where you install those pointers in the RTOC ( my
prefered way :-), but a more complex one. you'll have to create a shared
library that will contain entries for those functions, and than replace in
the TOC those fake entries with those is the 68K CR. This ensures that no
code change is required, and one can easily switch to a native version of
this library when it becomes available without changing the client, just
replacing the shared library )

2) build one mixed mode entry for the CR's main, and dispatch yourself.
This will work for register dispatched code.

Take a look at one of MPW's examples :

...:MPW:Examples:PPCExamples:CExamples:PPC-68K:

--
-- Etay Bogner,
-- [e-mail],
-- Scitex Corp.
-- Israel.

---------------------------

>From [e-mail] (Dave Fernandes)
Subject: How do I create many menus from one MENU resource?
Date: Fri, 19 Jan 1996 16:35:25 -0500
Organization: Marine Biological Laboratory

My problem is that I have a document window with a popup menu that is
filled in dynamically depending on the contents of the document.  If I
want to open more than one document, the popup menus in different document
windows might have different contents.  Do I need to have a seperate MENU
resource for each document I want to open?  Or is there a way to create
independent menus from one resource?

I'm doing this within PowerPlant, though I don't think that matters.

--
Dave Fernandes
Postdoctoral Research Associate
The Ecosystems Center
Marine Biological Laboratory

+++++++++++++++++++++++++++

>From [e-mail] (Clinton Popetz)
Date: 20 Jan 1996 00:51:11 GMT
Organization: Metrowerks Inc.

In article <davef-1901961635250001@128.128.173.90>, [e-mail] (Dave
Fernandes) wrote:

>My problem is that I have a document window with a popup menu that is
>filled in dynamically depending on the contents of the document.  If I
>want to open more than one document, the popup menus in different document
>windows might have different contents.  Do I need to have a seperate MENU
>resource for each document I want to open?  Or is there a way to create
>independent menus from one resource?
>
>I'm doing this within PowerPlant, though I don't think that matters.

This is the bane of popup menus.  The problem is that LStdPopupMenu (and
in fact most popup menu implementations since System 7) use the CDEF which
the system provides for Popups.  It keeps you from having to do things
like dynamically resize the frame, draw the drop shadow, etc. by hand.

The problem is that it expects a MENU resource number in the NewControl
call.  And if you have two popups loaded from the same resource, they will
effect each other when items are added/removed.

There are two ways people attempt to solve this.  Many programs just keep
a ton of empty MENU resources around, and hope they don't run out.  This
only delays problems in most cases.  (NCSA Mosaic kept 50 popups around.
But one day, someone hit a page where there were 51 forms, and WHAMMO)

There is another possibility that works for me, although no one has
verified that it is OK by the System.

I subclass LStdPopupMenu to make a CPopupWithDummy.   And I override
FinishCreateSelf in the following way:

void
CPopupWithDummy::FinishCreateSelf()
{
   //here be dragons

   MenuHandle newHandle = ::NewMenu(sMenuID++,"\p");
   ::InsertMenu(newHandle,-1);

   PopupPrivateDataHandle  dataH =
      (PopupPrivateDataHandle) (**GetMacControl()).contrlData;
   (*dataH)->mHandle = newHandle;
   (*dataH)->mID = mMenuID;
}

So I fool the CDEF into using a new MenuHandle.

sMenuID is a static that counts up, so that I don't use the same ID
twice.  Eventually this will turn over, and things will be bad.

But this WORKS, and I haven't gotten it to crash.  When the control is
destroyed, the MacOS CDEF will not call ReleaseResource on the Menu,
because it doesn't know how many times you are using it.

The other alternative is to do it the old System 6 way, where you use your
own CDEF and take care of everything by hand.  This is not fun...

                                       -Clint

--
/**
 * Clinton Popetz
 * [e-mail]
 * Arnold's Coffee Boy
 */

+++++++++++++++++++++++++++

>From [e-mail] (Jan Bruyndonckx)
Date: Mon, 22 Jan 1996 09:27:15 +0100
Organization: Cratylus Corp

In article <davef-1901961635250001@128.128.173.90>, [e-mail] (Dave
Fernandes) wrote:

> My problem is that I have a document window with a popup menu that is
> filled in dynamically depending on the contents of the document.  If I
> want to open more than one document, the popup menus in different document
> windows might have different contents.  Do I need to have a seperate MENU
> resource for each document I want to open?  Or is there a way to create
> independent menus from one resource?
>
> I'm doing this within PowerPlant, though I don't think that matters.

You can have several PP windows referencing the same popup menu, if the
menu contents stay the same.
In my program, the popup menu is hierarchical, and its contents are built
on the fly.  I use the NewMenu toolbox call for this, and I had to use a
different menuID for each one.


Jan                                  \|/
                                    (o o)
________________________________oOo__(_)__oOo_________________________________
|  Jan Bruyndonckx                                                           |
|  e-mail: [e-mail]                                     |
|  ALink:  WAVE.BEL                        "Small problems have deep roots." |
l____________________________________________________________________________l

---------------------------

>From [e-mail] (Andy Bettis)
Subject: MacApp newbie questions
Date: Mon, 22 Jan 1996 12:19:23 GMT
Organization: Insignia Solutions

I'm new to MacApp and am trying to learn it by the home study method,
however I'm hitting a few problems early on and I'd be grateful for some
expert advice.

First off, after building my app it refuses to run, claiming that the
required resources are not present. Is there a way of finding out what is
missing? I've compared by generated resource fork with some of the working
examples without spotting any obvious difficulties.

When constructing views with ViewEdit I can't add pop-up menus without a
corresponding MENU resource in addition to the CMNU (it crashes horribly).
And do I need to add pop-up menus to the hierarchic MBAR resource?

Can I directly change the fNumRows field in a TGridView subclass or do I
need to  insert the number of rows I need one at a time? And how is a
double-click selection passed to the app?

Finally, can anyone suggest a good 'how to' book? I'm OK on the C++ side
of things, and I have the Apple programmer's guide and tutorial manuals.

Thanks in advance, and could you mail any replies as I'm not a regular in
this group. Oh, by the way, I'm using CodeWarrior rather than MPW.

Rev. Andy

+++++++++++++++++++++++++++

>From [e-mail] (Trygve Isaacson)
Date: Tue, 23 Jan 1996 06:53:57 GMT
Organization: Freak Accident Music

In article <[e-mail]>,
[e-mail] (Andy Bettis) wrote:

> I'm new to MacApp and am trying to learn it by the home study method,
> however I'm hitting a few problems early on and I'd be grateful for some
> expert advice.
>
> First off, after building my app it refuses to run, claiming that the
> required resources are not present. Is there a way of finding out what is
> missing? I've compared by generated resource fork with some of the working
> examples without spotting any obvious difficulties.

This is commonly encountered by not including or initializing some of the
MacApp managers needed at startup. Usually it's the UDialog unit or its
resources.

So first, make sure you make these calls (you may not need them all, but
this basically covers it) in your main before you create and run the
application object:

   InitUMacApp(8);
   InitUPrinting();
   InitUTEView();
   InitUDialog();
   InitUMailer();
#if qDrag
   if (HasDragManager())
      InitUDragManager();
#endif
   InitUGridView();
   InitUSynchScroller();
   InitUFloatWindow();

Second, make sure you include the necessary resources by putting the
following lines in your main .r file (again, you may be able to omit some
of these, but start with them all to be sure):

#if qDebug
include "Debug.rsrc" not 'ckid';
#endif

include "MacApp.rsrc" not 'ckid';
include "Printing.rsrc" not 'ckid';
include "Dialog.rsrc" not 'ckid';
include "Defaults.rsrc" 'aete';
include "Defaults.rsrc" 'CLSS';

// Include 68K Code Resources when Rezzing
#if !qNoCode
include $$Shell("ObjApp")$$Shell("XAppName") 'CODE';
#endif

>
> When constructing views with ViewEdit I can't add pop-up menus without a
> corresponding MENU resource in addition to the CMNU (it crashes horribly).
> And do I need to add pop-up menus to the hierarchic MBAR resource?

Yes, put your popup CMNU ids in the hierarchical MBAR. You don't need any
MENUs, although you can use them where you don't need command number
mapping.

> Can I directly change the fNumRows field in a TGridView subclass or do I
> need to  insert the number of rows I need one at a time? And how is a
> double-click selection passed to the app?

You need to call TGridView methods such as InsRowBefore or InsRowFirst or
InsRowLast. These all allow you to insert several rows with one call. You
can't just alter fNumRows.

You need to handle double-clicks by overriding DoMouseCommand and checking
the event's click count.

Good luck...
...........................................................................
Trygve Isaacson                   "We secretly replaced this gourmet coffee
[e-mail]                    with Folger's Dilithium Crystals."
www.kagi.com/authors/trygve/
...........................................................................

+++++++++++++++++++++++++++

>From Lev Knopov < [e-mail]>
Date: 23 Jan 1996 07:17:51 GMT
Organization: Scitex Corp. Ltd.

[e-mail] (Andy Bettis) wrote:
>I'm new to MacApp and am trying to learn it by the home study method,
>however I'm hitting a few problems early on and I'd be grateful for some
>expert advice.
>
>First off, after building my app it refuses to run, claiming that the
>required resources are not present. Is there a way of finding out what is
>missing? I've compared by generated resource fork with some of the working
>examples without spotting any obvious difficulties.
>
>When constructing views with ViewEdit I can't add pop-up menus without a
>corresponding MENU resource in addition to the CMNU (it crashes horribly).
>And do I need to add pop-up menus to the hierarchic MBAR resource?
>
>Can I directly change the fNumRows field in a TGridView subclass or do I
>need to  insert the number of rows I need one at a time? And how is a
>double-click selection passed to the app?
>
>Finally, can anyone suggest a good 'how to' book? I'm OK on the C++ side
>of things, and I have the Apple programmer's guide and tutorial manuals.
>
>Thanks in advance, and could you mail any replies as I'm not a regular in
>this group. Oh, by the way, I'm using CodeWarrior rather than MPW.
>
>Rev. Andy

Dear Andy,

First of all, there is a MacApp dedicated newsgroup : comp.sys.mac.oop.macapp3.
Now , your questions :
1. MacApp is very depended on its resources. You need to include in your project
MacApp.r,DIalog.r,Debug,r(for Debug version),Printing.r as minimum. And of
course, you own resource build in ViewEdit. Do not forget to call
MA_REGISTER_CLASS(in 3.3) or macroDontDeadStrip(in 3.1.4) for views you
defined in ViewEdit. The only way to find out what is missing - debugger.
2. ViewEdit is not a best choice for MacApp views editing. Use Ad Lib
instead. Yuo donnot need to duplicate menus in CMNU and MENU.
3.You need to insert num of rows you need.
4.Double-click does not supported in TGridView, but you can override
DoMouseCommand of you view and examine TTolboxEvent for number of clicks.
5.Try to look on Apple MacApp examples - DemoText,DemoDialogs - it's a
good examples of MacApp appplications.
6. CodeWarrior is, to my mind, the best choice. Use Rez compiler to
compile MacApp resources.

Hope this helps

--
  //\\   Lev Knopov
  \\//   Scitex Israel
///  \\\ [e-mail]


Lev Knopov

[e-mail]
Scitex Israel



---------------------------

>From [e-mail] (Michael K. Neylon)
Subject: Matching struct definitions to TMPL resources
Date: 22 Jan 1996 14:21:11 GMT
Organization: University of Michigan Engineering, Ann Arbor

I have a preferences structure in my program, which I read in from
the data fork of my prefs file.  When generating this prefs file for
a new user, I have a 'pref' resource and a 'TMPL' resource.
The latter should describe the order and size of each item in the
'pref' resource, and then I set up the default values in the 'pref'
resource.

However, I am having trouble synchonizing the prefs TMPL to the
prefs structure.  One of the problems is that I cannot find
anywhere where there is a full list of 4-letter codes that describe
the types that can be used within the TMPL.  Also, I would think that
I may need to do some struct alignment within my program, but again,
I'm not sure if this is necessary, much less how to do it.

Any help with this would be appriciated...TIA...



--
 Michael K. Neylon, Graduate Student           | "Pausing only to reconstruct
 Dept. of ChE, Univ. of Michigan               |   the whole of integral
 [e-mail]                       |   mathematics, he continues
 http://www-personal.engin.umich.edu/~mneylon/ |   on..."    Marvin - tHGttG

+++++++++++++++++++++++++++

>From [e-mail] (Jeff Miller)
Date: Thu, 25 Jan 1996 19:59:04 -0800
Organization: Network TeleSystems, Inc.

In article <4e06gn$[e-mail]>, [e-mail]
(Michael K. Neylon) wrote:

> However, I am having trouble synchonizing the prefs TMPL to the
> prefs structure.  One of the problems is that I cannot find
> anywhere where there is a full list of 4-letter codes that describe
> the types that can be used within the TMPL.

>From the ResEdit Reference book (I apologize for the formatting, I copied
this from Acrobat):

These are the types you may choose from for your editable data fields:
DBYT, DWRD, DLNG Decimal byte, decimal word, decimal long word
HBYT, HWRD, HLNG Hex byte, hex word, hex long word
AWRD, ALNG Word align, long align
FBYT, FWRD, FLNG Byte fill, word fill, long fill (with 0)
HEXD Hex dump of remaining bytes in resource (This can only be the last
type in a resource.)
PSTR Pascal string (length byte followed by the characters)
LSTR Long string (length long followed by the characters)
WSTR Same as LSTR, but a word rather than a long word
ESTR, OSTR Pascal string padded to even or odd length (needed for DITL
resources)
CSTR C string (characters followed by a null)
ECST, OCST Even-padded C string, or odd-padded C string (padded with nulls)
BOOL Boolean (two bytes)
Note
Scrolling can become extremely slow if a template contains many BBIT or
BOOL items.
ResEdit does the appropriate type checking for you when you put the
editing dialog window away.
The template mechanism is flexible enough to describe a repeating sequence
of items within a resource, as in 'STR#', 'DITL', and .'MENU' resources.
You can also have repeating sequences within repeating sequences, as in
'BNDL' resources. To terminate a repeating sequence, put the appropriate
code in the template as follows: BBIT Binary bit (There must be 8 or an
even multiple of 8 of these; if fewer than 8 bits are defined, you must
include placeholder bits.)
TNAM Type name (four characters, like OSType and ResType)
CHAR A single character
RECT An 8-byte rectangle
Hnnn A 3-digit hex number; displays nnn bytes in hex format
Cnnn A C string that is nnn hex bytes long (The last byte is always a 0,
so the string itself occupies the first nnn-1 bytes.)
P0nn A Pascal string that is nn hex bytes long (The length byte is not
included in nn, so the string occupies the entire specified length.)
LSTZ
LSTE List Zero–List End. Terminated by a 0 byte (as in 'MENU' resources).
ZCNT
LSTC
LSTE Zero Count/List Count–List End. Terminated by a zero-based word count
that starts the sequence (as in 'DITL' resources).
OCNT
LSTC
LSTE One Count/List Count–List End. Terminated by a one-based word count
that starts the sequence (as in 'STR#' resources).
LSTB
LSTE List Begin–List End. Ends at the end of the resource. (As in 'acur'
and 'APPL' resources.)
The LSTB (list-begin) code begins the repeating sequence of items, and the
LSTE code is the end. Labels for these codes are usually set to the string
"*****". Both of these codes are required. It is generally advisable to
keep the beginning and ending labels identical to each other and to have
them be no more than five characters long.

Jeff

- -----------------------------------------------------------
 Jeff Miller  |  [e-mail]  |  Network TeleSystems, Inc.
- -----------------------------------------------------------

---------------------------

>From [e-mail] (mrw)
Subject: Menu Manager limitations?
Date: 20 Jan 1996 21:30:17 GMT
Organization: Johns Hopkins University

I can't find this information in IM.

What is the limit to the number of items I can put in a 'MENU'
resource?  I am displaying these items in a PowerPlant StdPopupmenu
object.

Thanks in advance,

-marty

+++++++++++++++++++++++++++

>From [e-mail] (John W. Baxter)
Date: Sat, 20 Jan 1996 20:52:30 -0800
Organization: Internet for the Olympic Peninsula

In article <4drmt9$[e-mail]>, [e-mail] (mrw) wrote:

>I can't find this information in IM.
>
>What is the limit to the number of items I can put in a 'MENU'
>resource?  I am displaying these items in a PowerPlant StdPopupmenu
>object.

You can put any number of items (any number you're likely to want to) into
a MENU resource as far as Inside Mac is concerned.  Only the first 31 can
be enabled/disabled in the resource or in code, since only those have bits
available in the 32-bit word in the resource and menu record which holds
those flags.

(I think there's a signed 2-byte count, so 32767 is the limit, but you
don't want that many, so it's moot.)

   --John

--
  "Officials said the collision was caused by the fog, which was as thick as sea poop."
John W. Baxter    Port Ludlow, WA     [e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Peter Castine)
Date: Wed, 24 Jan 1996 11:05:58 +0100
Organization: Process Control Center, TU Berlin

In article <4drmt9$[e-mail]>, [e-mail] (mrw) wrote:

>What is the limit to the number of items I can put in a 'MENU'
>resource?  I am displaying these items in a PowerPlant StdPopupmenu
>object.

You will take serious performance hits (not to mention having created an
abysmal user interface) long before you get to the technical limits of the
number of items in a menu.

There is a HIG recommendation to the effect that (with the exception of
dynamic menus, such as a Font menu) you shouldn't have more menu items
than what would fit _without_ scrolling on a 9" classic screen. This was
something like 17 or 21 items. Macs with 9" screens haven't been produced
for quite a while now, but there are still some around (I believe John
Norstrad's most valued UI-tester uses a Mac with 512x342 screen
resolution).

As to the technical limits: a look at the MenuInfo data structure suggests
the only limit is available memory; the Resource Manager has supported
resources > 32,767 bytes for years now. The implementation of Menu Manager
traps might put a limit at 32 or 16k items, but you'll have to do some ROM
disassembly to find out for sure (and even then, you never know what
future trap patches might do).

Hope this helps,

Peter

- -------- For a good time, http://www.prz.tu-berlin.de/~pcastine ---------
Dr. Peter Castine           | Whenever I hear anyone arguing for slavery, I
[e-mail]   | feel a strong impulse to see it tried on him
Process Control Center      | personally.
Technical University Berlin |                            -- Abraham Lincoln

---------------------------

>From [e-mail] (hypervision)
Subject: Programming a PinnBall
Date: Tue, 02 Jan 1996 22:45:11 -0100
Organization: hypervision

A couple of days ago i finished my first (beta) pinball.
Collision detection, bouncing, gravity, multiple heigths, whatever... it
all works nice and fast. The only problem is that i have some problem with
the 'feel' of the game : the ball doesn't seem to make a really realistic
drop (although the algorithm is pure physics, and it is correct (i hope))

I guess the trick is to set the parameters correctly. So here i go :

   - when calculating gravity, i add (every 5 ticks or so) a small amount
to the ball's vertical velocity.  how much should i add ? (in terms of the
ball's size, which is variable and currently set to 40x40 pixs in a
640x480 pinball arena)
I tried a bunch of values ranging from 0.5 to 1.5, but they all seem weird
(greater than 1.5 flips the ball out of the arena)

   - when the ball bounces an object, i calc the refraction vector and set
that to the new velocity of the ball. Should i scale this velocity also ?
How much would be a logical value (considering non-bouncing objects. i.e.
plain walls)
   - am i forgetting something crucial ?


If i ever finish this thing entirely (currenlty working on sounds and
moveable parts), i'll probably release the source code.

Thanks in advance

Grendel.

+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Tue, 02 Jan 1996 17:49:55 -0800
Organization: Qualcomm, Inc.

In article <hvision-0201962245110001@193.75.128.7>,
[e-mail] (hypervision) wrote:

Re: getting a pinball simulator to simulate well:
>    - when calculating gravity, i add (every 5 ticks or so) a small amount
> to the ball's vertical velocity.  how much should i add ? (in terms of the
> ball's size, which is variable and currently set to 40x40 pixs in a
> 640x480 pinball arena)
> I tried a bunch of values ranging from 0.5 to 1.5, but they all seem weird
> (greater than 1.5 flips the ball out of the arena)
>
Well, I have no idea what kind of value is appropriate numerically, but
you can take some physical measurements and do some back-of-the-envelope
calculations to come up with a reasonable number.  Here's some tips to get
you started...

First of all, it doesn't matter if your ball is huge or not.  From
physics, you know F=ma, and the acceleration due to gravity is a constant
9.8 m/s^2.  Now your ball isn't moving vertically through space, so you
know that it can't be effectively feeling the full force of gravity.
Given the length of the pinball table and the height at the scoreboard
(both of which are easy to measure) you can compute the angle of tilt on
the pinball table using trigonometry.  Multiply the 9.8 m/s^2 for vertical
motion times the sine of the angle of tilt, and you'll have the effective
acceleration on the ball.

This number is in m/s^2, so you need to start converting units to get
things right.  Find an appropriate conversion factor between physical
lengths and pixel distances, and some constant time multiplier to get your
times right, and you should be all set.  You will have to tweak these
conversion factors to get the effect you want, but once you're all done,
it should hopefully look pretty good.

Of course, this is physics-dork way of doing things, and you might want to
just play with numbers instead (as you were doing) until something looks
cool.

>    - when the ball bounces an object, i calc the refraction vector and set
> that to the new velocity of the ball. Should i scale this velocity also ?
> How much would be a logical value (considering non-bouncing objects. i.e.
> plain walls)

You'd calculate the reflected velocity vector, and set the ball's new
velocity to that.  Realistically, you should scale this number by some
constant factor to account for the elasticity of the collision.  TOTMGPG
has some stuff on this, but the gist is this: if you multiply by an
elastic factor of 0, then your balls stops when it hits the object.  This
is an inelastic collision, and you definitely don't want that.  If you set
the elasticity to 1, then you ball doesn't lose any energy in the
collision, and it has the same speed entering and leaving.  Realistically,
you want your ball to lose a bit in the collision so try 0.8 or so for
that factor.

>    - am i forgetting something crucial ?
>
Just remember to be somewhat accurate in your calculations.  Nothing is
more annoying than a pinball game where it's REALLY easy to make the same
shot twice.

Good luck!
Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 3 Jan 1996 03:36:30 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> [e-mail] (Jason Regier) writes:
>In article <hvision-0201962245110001@193.75.128.7>,
>[e-mail] (hypervision) wrote:
>>    - am i forgetting something crucial ?

>Just remember to be somewhat accurate in your calculations.  Nothing is
>more annoying than a pinball game where it's REALLY easy to make the same
>shot twice.

For a good game of pinball, I would think that it would be important to
model ball spin as well. It's mathematically a bit tougher than the
collisions and motion without it, but it should be possible to take
shortcuts. For a really good implementation, I would think that
quaternions might come in handy.

I don't know if any computer pinball games model spin...all of the ones
that I have tried have felt more like video games rather than the real thing.

It may also be a good idea to run the simulation at an extremely high frame
rate (say 100-500 frames per second) and draw the results at around the
screen refresh rate (motion blurring the ball is a very good idea).

The material of the bumper should determine how the ball bounces off on
a collision. A very hard surface will conserve energy very well. A soft
surface will absorb the energy. A rubbery surface would first absorb the
energy and then bounce back, restoring the momentum (but not the spin).

Most computer pinballs seem to use brownian motion simulation to jitter
the ball so that it doesn't get stuck so easily. I don't like it when
they overdo it, but I guess it's a necessary evil.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (John W. Blackburne)
Date: Wed, 03 Jan 1996 21:28:36 +0800
Organization: Tempest

In article <hvision-0201962245110001@193.75.128.7>,
[e-mail] (hypervision) wrote:

:A couple of days ago i finished my first (beta) pinball.
:Collision detection, bouncing, gravity, multiple heigths, whatever... it
:all works nice and fast. The only problem is that i have some problem with
:the 'feel' of the game : the ball doesn't seem to make a really realistic
:drop (although the algorithm is pure physics, and it is correct (i hope))
:
:I guess the trick is to set the parameters correctly. So here i go :
:
:   - when calculating gravity, i add (every 5 ticks or so) a small amount
:to the ball's vertical velocity.  how much should i add ? (in terms of the
:ball's size, which is variable and currently set to 40x40 pixs in a
:640x480 pinball arena)
:I tried a bunch of values ranging from 0.5 to 1.5, but they all seem weird
:(greater than 1.5 flips the ball out of the arena)

First for realism you should have some sort of constant time base. One way
is to do your calcs at frequent fixed intervals, e.g. 20 times a second,
either by making sure your graphics can be done in 0.05 sec then waiting
between redraws until 0.05 sec later, or by having a time manager task
doing the velocity calcs independent of screen redraw every 0.05 sec.

If you find it difficult to do your calcualations at fixed times use the
time between your calculations in your calcuations as a scaling factor.
E.g. if your accn is m then use

       this_time = MyGetTime()
       v = v + m * (this_time - last_time)
       last_time = this_time

In the above m is acceleration due to gravity. On a sloped surface this
should be about 9.8 * sin (angle_of_slope) (m/s^2), scaled to fit it on
screen. (Both v and m are vectors, so the above is not real code in any
language I use.)

For added realism you could include a drag factor: this is probably not
very significant in pinball but it does act as a limiting factor on speed,
and so is noticable in some circumstances. Adding in a drag factor d to
the above formula you get

       v = v + (m - (d * v)) * (this_time - last_time)

The best value for d can be determined by trial an error, but it should
probably be a small number, e.g. 0.1. One guide is that the above equation
gives no change in velocity when (m - (d * v)) = 0, i.e. when v = m/d,
which will be the 'terminal' or maximum velocity of the ball as it rolls
down the sloped table.

You should do your calcuations at a much higher resolution than the screen
resolution, e.g. use fixed point or floating point arithmetic. Fixed point
is best where you're doing a lot of calcuations and can't rely on a
floating point co-porcessor being present, though with only one moving
object most of the time you might find floats easier. Then just take the
integer part when plotting to screen.

:   - when the ball bounces an object, i calc the refraction vector and set
:that to the new velocity of the ball. Should i scale this velocity also ?
:How much would be a logical value (considering non-bouncing objects. i.e.
:plain walls)

The easiset way is probably to use surface normals, then do something like

      v = v - n * ((v.n) * (1 + e))

n is the unit normal vector, the '.' is a dot product, and e is the
coefficient of resitution. You then just need a table of n and e for each
surface. As Juri noted the amount of reflection will vary for different
surfaces, from e=0 for no bounce to e=1 for a perfect bounce, with no
energy lost. Again experiment to see what is best.

You might also want to include spin, but this makes it a fair bit more
complex, as at each colision you'd want to not only check how much spin
effected the impact (typically by ginving it a boost along the surface)
but also how much the surface affected he spin, which would depend on the
surface and the angle of impact. Spin would also effect it's trajectory.

:   - am i forgetting something crucial ?

Magnets. Some of the most fun pinball games I've played have had magnets,
or even better electro-magnets, invisibly embedded in the playing surface.
Note that I am not a very good pinball player, so the degree of randomness
such would introduce did not bother me.

Tilt. Not sure how you'd do this, I've seen a couple of computer
implementations, neither of them impressed me.

Pits/ramps. You could use most of the calculations above but would want to
vary both the magnitude and direction of the vector m.

John
--
John Blackburne,                     [e-mail]
Programmer  Asia, Inc. Online:       http://www.asia-inc.com
Technology consultant and trainer:   http://www.hk.super.net/~johnb

+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Wed, 03 Jan 1996 10:15:14 -0800
Organization: Qualcomm, Inc.

In article <johnb-0301962128360001@0.0.0.0>, [e-mail] (John W.
Blackburne) wrote:

> Adding in a drag factor d to the above formula you get
>
>        v = v + (m - (d * v)) * (this_time - last_time)
>
I think you might get more "realistic" results if you tried d * v^2 and
made d a small number.  If you make d too high, it will look like playing
pinball in a viscous medium like oil, which is a pretty strange effect.
Honestly, you might want to scrap drag altogether.  The drag on a pinball
is pretty darn small-- most of your energetic losses occur when you hit
objects like bumpers and targets.

Juri, you brought up spin.  Spin would be REALLY cool to see implemented.
But are you sure that you'd want to use quaternions?  Quaternions are
great for doing rotations in 3D space, but for a top-down pinball game
(like all the ones I've seen) you don't have 3 angles of rotation to deal
with.  There should be only 2: one in the vertical plane, and one in the
horizontal plane (horizontal meaning in the plane of the screen, and
vertical meaning perpendicular to the screen).  If you further assume that
all your bumpers and flippers make contact with the ball through the
center of the ball, then you can't really impart any weird spin in the
vertical direction, so the problem pretty much reduces down to one angle.
You can calculate the amount of horizontal spin by simply looking at where
the ball contacts an object, the rotational moment of inertia for a
sphere, and some physics.

> Magnets.
Way cool, and probably pretty easy to add.  Just assume some 1/r^2 force
relation for your "point" magnetic source (where r is the distance between
the ball and the magnet).  This just appears as another acceleration
acting on your ball when you go to calculate the ball's next position and
velocity vectors.  And when the magnet's not on, the force/acceleration
disappears.

> Tilt. Not sure how you'd do this, I've seen a couple of computer
> implementations, neither of them impressed me.
>
I'd guess that you check to see if you're within some distance of an
object.  If you are, then make the ball hit the wall.  Tilting and
slapping is more like imparting an impulse to the ball, which corresponds
to an instantaneous change in velocity.  If you are slamming the pinball
table forward (i.e. vertically on the screen), then when calculate the
collision, SUBTRACT some number from the ball's vertical velocity when
calculating the collision.  Likewise, if you slap on the left of the
table, when you do the collision, make the ball appear as if it were
really traveling faster towards the left than it really is.

> Pits/ramps. You could use most of the calculations above but would want to
> vary both the magnitude and direction of the vector m.
>
For pits and ramps, you could probably get away with faking it.  It seems
pretty silly to me to make all your calculations in 3D just because you
add one ramp.  Here's how I'd fake a ramp:  I basically am envisioning a
two-state system, where you're either on the ramp or you're not... let a
boolean variable RAMP (one for each ball) keep track of this.  As the ball
passes through the entrance to a ramp, toggle RAMP appropriately.  Thus,
if you never pass through the entrance, RAMP is 0, indicating that you
never made it up the ramp, and your game progresses normally.  If RAMP is
1, then you're on the ramp.  Draw appropriately, and add forces when on
the ramp to simulate the hump that you have to get over to stay on the
ramp.

This a pretty cool discussion!

Later,
Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 4 Jan 1996 02:30:28 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> [e-mail] (Jason Regier) writes:
>Juri, you brought up spin.  Spin would be REALLY cool to see implemented.
>But are you sure that you'd want to use quaternions?  Quaternions are
>great for doing rotations in 3D space, but for a top-down pinball game
>(like all the ones I've seen) you don't have 3 angles of rotation to deal
>with.  There should be only 2: one in the vertical plane, and one in the
>horizontal plane (horizontal meaning in the plane of the screen, and
>vertical meaning perpendicular to the screen).  If you further assume that
>all your bumpers and flippers make contact with the ball through the
>center of the ball, then you can't really impart any weird spin in the
>vertical direction, so the problem pretty much reduces down to one angle.
>You can calculate the amount of horizontal spin by simply looking at where
>the ball contacts an object, the rotational moment of inertia for a
>sphere, and some physics.

I didn't give it too much thought, so maybe quaternions are overkill,
but it seemed to me that spin would affect the ball motion when it
was traveling (touching only at the bottom), so you would want to
know the distance and direction it has rotated at the bottom and
bleed some of that into the acceleration or deceleration fo the ball.
This also goes the other way: some of the ball speed is converted
into rotational energy. If the spin of the ball is in synch with
the motion, no energy transfer happens.

Then, when the ball hits an obstacle, it might be necessary to get the
rotation in the same manner, but against the surface that it hit.

Spin is one of the reasons why the ball may fly up when it hits an
obstacle: the ball rotation is converted into upward motion.

I think rotational physics are my weakest area, so you may be right
and I might be wrong. Still, I'm sure quaternions would guarantee
the correct results whereas taking shortcuts could produce more
complicated code that doesn't quite match reality.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Thu, 04 Jan 1996 10:05:23 -0800
Organization: Qualcomm, Inc.

In article <4cfe44$[e-mail]>, [e-mail] (Juri Munkki) wrote:
> I didn't give it too much thought, so maybe quaternions are overkill,
> but it seemed to me that spin would affect the ball motion when it
> was traveling (touching only at the bottom), so you would want to
> know the distance and direction it has rotated at the bottom and
> bleed some of that into the acceleration or deceleration fo the ball.
Since we're kinda assuming that there is negligible friction between the
ball and the waxed pinball table, then the spin of the pinball won't
affect the way the ball moves across the surface of the table.  If you
assume that there is some friction, then you're right, the spin of the
ball will affect its motion across the table.

But regardless of table friction, the spin of the ball WILL affect its
path after a collision.  And like you said, this would produce a variety
of neat, strange effects.  Given the rotational and translational
velocities of the ball prior to the collision, you can find it's starting
energy.  If the collision is say, 80% elastic, you know that your energy
after the collision is %80 of the starting energy.  The trick is
determining how much of that final energy goes into translational motion
and how much goes into rotational motion.

Oh, and I just think quaternions seem like overkill because they're
designed for rotations in 3D space.  If you take the top-down view, then
like I said before, the only angle you really care about is rotation in
the plane of the table.

> Still, I'm sure quaternions would guarantee the correct results whereas
taking > shortcuts could produce more complicated code that doesn't quite
match
> reality.
>
Yep, you may be right.

Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 5 Jan 1996 01:13:50 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> [e-mail] (Jason Regier) writes:
>Since we're kinda assuming that there is negligible friction between the
>ball and the waxed pinball table, then the spin of the pinball won't
>affect the way the ball moves across the surface of the table.  If you
>assume that there is some friction, then you're right, the spin of the
>ball will affect its motion across the table.

I would think that the friction between the ball and table can be
considerable, if the ball is not rolling on the surface. If the spin
is left out and the friction with the the table is left out, you
don't have a pinball game: you get a weird sort of air hockey or
"shufflepuck" game.

If you want to see ball spin affecting its trajectory on a table,
take a look at the Williams pinball Whirlwind. The ball acts like
it is drunk after a few spins in the whirlwind.

My advice to the person who was writing this pinball is: learn the
physics, do the math and impress everyone by the incredible realism.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From Robert Grant <[e-mail]>
Date: Fri, 05 Jan 1996 13:27:39 -0800
Organization: Informix Software, Inc. Menlo Park, CA 94025

All this spin stuff is interesting, but how is the interaction
with the flipper handled? Something that is very common in
pinball is catching the ball with the flipper and then letting
it roll down before flipping it back up the table - hopefully
at the desired angle. This seems to me to be fairly
complicated or am I missing something?

Robert.

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 6 Jan 1996 03:18:37 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> Robert Grant <[e-mail]> writes:
>All this spin stuff is interesting, but how is the interaction
>with the flipper handled? Something that is very common in
>pinball is catching the ball with the flipper and then letting
>it roll down before flipping it back up the table - hopefully
>at the desired angle. This seems to me to be fairly
>complicated or am I missing something?

Laws of physics are not really complicated. Yet, if you try to take
shortcuts, the shortcuts can become complicated.

A very high internal frame rate combined with rotational energy
calculations should make the flipper interaction realistic.

The roll of the ball keeps it glued to the flipper and guide.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (Samuel Zehr)
Date: Sat, 06 Jan 1996 02:09:18 -0600
Organization: Pixie Enterprises

In article <4chu0e$[e-mail]>, [e-mail] (Juri Munkki) wrote:

> In article <[e-mail]>
[e-mail] (Jason Regier) writes:
> >Since we're kinda assuming that there is negligible friction between the
> >ball and the waxed pinball table, then the spin of the pinball won't
> >affect the way the ball moves across the surface of the table.  If you
> >assume that there is some friction, then you're right, the spin of the
> >ball will affect its motion across the table.
>
> I would think that the friction between the ball and table can be
> considerable, if the ball is not rolling on the surface. If the spin
> is left out and the friction with the the table is left out, you
> don't have a pinball game: you get a weird sort of air hockey or
> "shufflepuck" game.
>
Actually, the ball doesn't roll as much as you would think - it glides.
That's what makes pinball so non-obvious. The ball is on a mylar surface,
so the spin of the ball can easily be opposite to the direction of
movement. In a factory new pinball game, the ball really shouldn't 'roll'
at all...
sz

--
Samuel Zehr               "My mom said it looked like a
Anonymous title            pretty little rose"
Anonymous company              --Bernd Meierbachtol

+++++++++++++++++++++++++++

>From [e-mail] (James Shaw)
Date: 11 Jan 1996 15:24:26 GMT
Organization: Computer Science Department, Princeton University

Jason Regier writes:
> This a pretty cool discussion!

Agreed.  I've also been working on a shapes engine to support pinball
as well as some other physically based games.  I've got the basic
collision detection for most important shapes in but I've only got the
typical simple rotationless physics.

I'd like to continue this discussion (either in news or in email)
of the solutions (and implementations) of spin and rotation.

Juri started it out with:
 > For a good game of pinball, I would think that it would be important
to
 > model ball spin as well. It's mathematically a bit tougher than the
 > collisions and motion without it, but it should be possible to take
 > shortcuts. For a really good implementation, I would think that
 > quaternions might come in handy.

And Jason continues with:
 > Juri, you brought up spin.  Spin would be REALLY cool to see
implemented.
 [ Discussion of quaternions deleted. ]

I'll agree with Juri; I think quaternions are needed.  There are three
angles of rotation:  spin (e.g. from top view the ball acts like a top),
rolling (the balls rotational motion in the direction it is moving) and
"english" or anti-rolling (the rotation part perpendicular to the balls
direction which takes it off course).

For example, from an unmoving ball getting struck by the plunger,
shouldn't the need to start rotating, bleed some translational
energy away.  And the big question: how does rolling bounce?

Also, even if quaternions aren't needed for pinball.  How about
a realistic billiards game?

Juri continued with:
 > If you want to see ball spin affecting its trajectory on a table,
 > take a look at the Williams pinball Whirlwind. The ball acts like
 > it is drunk after a few spins in the whirlwind.

 > My advice to the person who was writing this pinball is: learn the
 > physics, do the math and impress everyone by the incredible realism.

Good suggestion.  I'd like to take you up on it but, better yet,
I'd like to get a small list of people who would like to discuss
this in detail.  I'll be happy to start with my guess of the physics
but I'd like others to discuss it with and to check my work.
I'm hoping that either Juri or Jason or both would be interested,
but also any lurkers who think that they can contribute.

Interested?  Please email me, or just continue the thread.

Best,
James
[e-mail]


+++++++++++++++++++++++++++

>From [e-mail] (Francois Pottier)
Date: 11 Jan 1996 19:54:53 GMT
Organization: Ecole Normale Superieure, Paris

In article <4d3a3a$[e-mail]>,
James Shaw <[e-mail]> wrote:

>I'll agree with Juri; I think quaternions are needed.  There are three
>angles of rotation:

How are quaternions used for modelling rotations? In classical
physics any rotation is usually represented by a vector W, and a
vector product is used to compute the image of a given vector
by the rotation, i.e. after a time t, vector X becomes
Wt /\ X.

(W is actually called a pseudo-vector because it is dependent on the
choice of a direct base, but that is a physical consideration which
isn't very important for modelling).

As far as I know quaternions are elements of a non-commutative
field H. How can they be used for rotations.

It's been a few years since I left the physics class...

--
Francois
[e-mail]
http://www.eleves.ens.fr:8080/home/pottier/

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 12 Jan 1996 11:09:56 GMT
Organization: Helsinki University of Technology

In article <4d3pud$[e-mail]> [e-mail] (Francois Pottier) writes:
>How are quaternions used for modelling rotations?

SIGGGraph'85, page 245: "Animating Rotations with Quaternion Curves" by
Ken Shoemake. It mostly deals with interpolating rotations using
quaternions, but provides a lot of references and a basic discussion
of how quaternions convert to rotation.

The Graphics Gems book series also has a lot of articles on quaternions.
I still only have the first book (shame on me), but it has a basic discussion
of quaternions on page 498: "Using Quaternions for Coding 3D Transformations"
by Patrick-Gilles Maillot. The source code is available on anonymous ftp.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/jmunkki                  Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Mon, 15 Jan 1996 16:36:03 -0800
Organization: Qualcomm, Inc.

In article <4d3a3a$[e-mail]>, [e-mail] (James
Shaw) wrote:
> I'll agree with Juri; I think quaternions are needed.  There are three
> angles of rotation:  spin (e.g. from top view the ball acts like a top),
> rolling (the balls rotational motion in the direction it is moving) and
> "english" or anti-rolling (the rotation part perpendicular to the balls
> direction which takes it off course).
>
First of all, quaternions are 4 component vectors with some really neat
properties that lend themselves quite nicely to rotations, especially in
3D graphical systems.  I don't know nearly as much about these as I could
or probably should, but for more info, check out more advanced 3D Computer
Graphics books.  Also, Embedded Systems magazine has had several extensive
articles in the past about rotations with quaternions, most of them called
something like "The Spin Doctors".  If I understand correctly, quaternions
basically give you some powerful alternatives to dealing with Euler
angles.

If you're dealing with 2D systems, I get the strong impression that
quaternions may be a little heavy duty for the job.  Physically, if you
apply a force through the center of mass of an object, you're imparting
purely translational energy and zero rotational energy.  But we all know
that frictional forces act at the edge of a ball, so there's bound to be
some energy going into the rotation of the pinball as it moves across the
playing surface.

But the question is can you see those 3 dimensions of spin on a pinball in
real life?  I can't.  I can see when somebody spins the ball really fast
(i.e. Whirlwind) and it dances around like a top.  But I've never seen a
pinball react to some backspin like a cue ball does on a pool table.
Thus, if I were modelling a pinball game, I'd ignore backspin and only
simulate the spin I could see in the real world.

> Also, even if quaternions aren't needed for pinball.  How about
> a realistic billiards game?
>
Well, that's a different story.  I don't think there's really any way you
can do a realistic billiards game without taking into account all the
directions of english.  (You might think there's only back/topspin and
side english, but masse and jump shots can produce spin in the 3rd
direction).  Sounds to me like quaternions might be really handy here.
And I'd love to see a great pool game-- I've yet to check out Virtual
Pool.  Anybody know if it's good?

Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (John W. Blackburne)
Date: Tue, 16 Jan 1996 19:12:45 +0800
Organization: Tempest

In article <[e-mail]>,
[e-mail] (Jason Regier) wrote:

:But the question is can you see those 3 dimensions of spin on a pinball in
:real life?  I can't.  I can see when somebody spins the ball really fast
:(i.e. Whirlwind) and it dances around like a top.  But I've never seen a
:pinball react to some backspin like a cue ball does on a pool table.
:Thus, if I were modelling a pinball game, I'd ignore backspin and only
:simulate the spin I could see in the real world.

I would forget about backspin on this too. Apart from it being invisible
it would not appear because of the dynamics. You can get sidespin from a
ball striking a rough barrier at a shallow angle. To get backspin you
would need to either have the ball contact a barrier on a point on the top
or bottom. You could do that with a barrier like this

                          _________
                          \
                   --------

 (or something even steeper) but how many of these do you see on a pinball
table ? Or you could 'flick' a ball off a barrier, as you can with a
marble by pushing down with it under the edge og your thumb, but again
does this mak sense on a pinball table.

I'm not even sure sidespin would be necessary. I recently downloaded the
demo of the Mac pinball game 'Loony Labyrinth', and while watching it run
in demo mode I carefully looked out for any sign of spin, and could see
none, as far as I could tell.

But it's a fine game. What makes it so ? A varied, interesting,
multi-level table, excellent graphics and sound and very smooth gameplay,
with very straighforward mechanics as far as I could tell.

John
--
John Blackburne,                     [e-mail]
Programmer  Asia, Inc. Online:       http://www.asia-inc.com
Technology consultant and trainer:   http://www.hk.super.net/~johnb

+++++++++++++++++++++++++++

>From [e-mail] (john calhoun)
Date: 18 Jan 1996 04:21:00 GMT
Organization: Scheherazade Software

In article <[e-mail]>, [e-mail] (James Shaw) wrote:

> This had an interesting effect on the simulation.  I had to
> define how elastic collisions worked for a ball striking a surface.
> I ruled that the spin reverses along the normal, just like translational
> motion.  It can be anywhere from inelastic to elastic just like
> translational collisions.  I have this image in my head of a rolling
> gear catching against the side of something and bouncing back, but
> otherwise, don't know if this bouncing is physically valid.  Comments?

Doesn't sound right.  If the ball is spinning like a top when it collides
with a barrier, you would expect the angle of deflection (off the barrier)
to be modified and (perhaps) the spin to completely stop at the instant of
collision (giving over it's rotational momentum to a change in direction
of the ball).  Imagine a billiard ball spinning like a top hitting a side
rail dead on (perpindicular), it shouldn't bounce straight back along the
path of incidence - it should instead come off at some angle.  For the
instant at which the side of the ball contacts the side rail (assuming
friction and thus elasticity) there will be a sudden torque that converts
into a translational (no longer rotational) force.  I think the velocity
the ball was going when it struck the rail (how hard it pressed into the
rail) and the the coefficent of friction between the ball and the rail
would determine both the maginitude of the torque moment and thus the
degree to which the spin was reduced.

> Rolling had the effect of, with gravity, losing energy in the system.
> This was at least in my implementation.   I had an explanation in
> my head why this would happen but I can't explain it at the moment.

I don't understand your wording on this for several reasons.  I don't see
how gravity causes a thig to "lose" energy, nor do I see how rollinf or
spin would either.  Now, if a ball has back soin that is slowly translated
into translational motion, I would expect some friction between the ball
and the surface and thus some heat produced.  Perhaps this is what you
meant by a loss of energy in the system.

> I didn't do anything with the interaction of the 3 directions of
> rotation.  I think they are independent but I don't have a physical
> arguement yet.

I don't buy into that exactly.  There is a definite "gyroscope" action
going on that will prevent a spinning top (or ball) from rolling over, but
friction with the surface the ball is on might tend to erode that.  Get a
ball spinning like a top on a coffee table, and then pick up one end of
the coffee table.  You expect the incline to start the ball sliding down
the table - and you expect as well the ball to try and maintain it's
original spin.  But how would the friction of sliding down the table act
on the ball?  Would that small moment of force start to tip the balls
spin?

Sounds like you might just need to tweak your system a bit and you'll have it.

john calhoun-

+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 18 Jan 1996 00:39:40 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> [e-mail] (James Shaw) writes:
>If anyone else is interested, I'd like to start an email discussion of
>the physics.  There are enough mistakes that can be made so multiple
>participants would be helpful.

I think the topic is very suitable for the newsgroup, so I would keep
the discussion alive here and let everyone contribute.

I'm extremely tied up with Avara, so any time spent here comes from
my recreational time quota and shouldn't involve "deep thought". ;)

I'm glad you tried the spin and I'm glad you like the results. Even
if it is not absolutely correct, the added twist of handling spin
(does that count as pun?) will probably be appreciated by people who
play your game.

It still seems to me that quaternions would provide the ideal simple
solution for handling all kinds of spin: with any kind of motion
you'll have just a surface normal for the surface that you are
hitting or touching, a motion vector and a quaternion. The tough
part is then handling the situation where the ball bounces up from
the 2D plane.

But as with anything else having to do with rotations, quaternions
are just one solution. I haven't used them much (just some rotation
interpolations for objects and camera motion in 3D), but I have a
feeling that they would be the elegant solution for pinballs as well.

As long as you have a limited amount of balls (say under 10), most
machines should be able to handle any kind of math you want to throw
at them.

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/jmunkki                  Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (Bo Lindbergh)
Date: 19 Jan 1996 03:06:53 GMT
Organization: Royal Institute of Technology, Stockholm, Sweden

In article <4dk4sc$[e-mail]> [e-mail] (Juri Munkki) writes:
>
> As long as you have a limited amount of balls (say under 10), most
> machines should be able to handle any kind of math you want to throw
> at them.

Hmm, what's the greatest number of balls simultaneously in play ever used
in a physical pinball machine?  And if I ask this in rec.games.pinball,
will they laugh at me? :-)


/Bo Lindbergh


+++++++++++++++++++++++++++

>From [e-mail] (James Shaw)
Date: Mon, 15 Jan 1996 16:07:07 GMT
Organization: Olsen & Associates AG, Zurich, Switzerland

I said earlier:
 > I'll agree with Juri; I think quaternions are needed.  There are three
 > angles of rotation:  spin (e.g. from top view the ball acts like a top),
 > rolling (the balls rotational motion in the direction it is moving) and
 > "english" or anti-rolling (the rotation part perpendicular to the balls
 > direction which takes it off course).

I take it back.  Quaternions are not needed.  I implemented rolling,
english and spinning this weekend and needed only a 3D vector to
store them.  Of course, if I wanted to display the rolling object's
absolute position, I would need a rotation matrix or quaternions,
but for the spin's velocity, I only needed one vector.

Here is what I did in my implementation:

I assumed that the ball could only be hit on the side and effectively
on the bottom.  I stored the rolling/english as the x, y portion
of my vector w, and defined the x movement to be the rotation direction
which moves the ball in the x direction; same for y.

If the ball spins with rate w and the ball is moving at velocity v,
then the ball is touching (hitting) the ground and there is an
apparent velocity between the ground and the ball of a = w - v.
(If there is no slipping and the ball rolls along, a = 0)

Because there is some (a lot) of slipping, for one time step, I
define f, the friction of the ball and the ground.  That means
that I add f * a to w and subtract it from v.  This makes them match.
I actually should be using a ratio of the mass of the object to
its moment, but I've just set it to 1 for now.

Summary for rolling:
        a = w - v;
        w += MtoM * f * a;
        v -= f * a;
This accomplishes converting translation into rolling and
vice versa.  It does it via slipping and catching according to f.

This had an interesting effect on the simulation.  I had to
define how elastic collisions worked for a ball striking a surface.
I ruled that the spin reverses along the normal, just like translational
motion.  It can be anywhere from inelastic to elastic just like
translational collisions.  I have this image in my head of a rolling
gear catching against the side of something and bouncing back, but
otherwise, don't know if this bouncing is physically valid.  Comments?

The rolling's effect was to dampen instantaneous translational motion.
If the ball hit an active bumper, it went quickly away but slowed down as
the ball slipped and both rolled faster and moved slower.  If I
turned off translational elastisity, then when the ball hit something
it would thud to a stop, but be spinning away and as it 'caught' the
floor it would start to take off, like a race car.  It looked good.

Rolling had the effect of, with gravity, losing energy in the system.
This was at least in my implementation.   I had an explanation in
my head why this would happen but I can't explain it at the moment.

I did a similar slipping for spin of the ball (the z portion of w)
and am still playing with the ball spinning along object, which from
the top view and gravity on is what you might call rolling.

I didn't do anything with the interaction of the 3 directions of
rotation.  I think they are independent but I don't have a physical
arguement yet.  I keep thinking that a spinning ball pushed ahead
would start to veer in the direction of the spin.   I'm also remembering
that a spin in one direction could resist spins in other directions?
I'm playing with this but I'd appreciate anyone out there who has an
opinion of the actual physics involved.
I haven't had a chance yet to put a whirlwind type rotater in
and see if the ball wobbles, as Juri mentioned.  My guess is that
it won't (yet?) because I'm missing something with the physics.

If anyone else is interested, I'd like to start an email discussion of
the physics.  There are enough mistakes that can be made so multiple
participants would be helpful.

Best,
James


+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Thu, 18 Jan 1996 10:13:57 -0800
Organization: Qualcomm, Inc.

In article <4dk4sc$[e-mail]>, [e-mail] (Juri Munkki) wrote:
> It still seems to me that quaternions would provide the ideal simple
> solution for handling all kinds of spin: with any kind of motion
> you'll have just a surface normal for the surface that you are
> hitting or touching, a motion vector and a quaternion. The tough
> part is then handling the situation where the ball bounces up from
> the 2D plane.
>
THAT would be really interesting to see.  I've never seen a pinball
simulator where you can actually hit the ball so hard that it smacks off
the glass (as in real pinball games).  Of course, in order to make your
pinball hop, you'd need to keep track of the heights of the bumpers and
other objects.  Well, at least you need to know the height at which it
will contact the ball (i.e. if bumpers contact below the ball's center of
mass, then you can hop the ball off the surface of the table!)

> As long as you have a limited amount of balls (say under 10),
This is a parameter related to the pinball game, and not the programmer,
right? ;)

Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Geoffrey Kinnel)
Date: 18 Jan 1996 19:10:24 GMT
Organization: Drexel University

This discussion of spin in pinball is interesting, but don't forget that
pinball tables are covered with mylar, and there is not a whole lot of
friction as the ball slides around the table. Compare this to a pool
table, where the surface is felt and you can use the spin. For more
pinball related stuff, check out rec.games.pinball and the Pinball Archive
http://pinball.cc.cmu.edu.

As an aside, many of the wierd effect people attribute to "playfield
magnets" are actually due to spin on the ball. I realize this sounds like
it is in contradiction with the low friction playfield; I don't pretend to
understand it.

I hope this helps someone, at least a little.

Geoff



+++++++++++++++++++++++++++

>From [e-mail] (Jason Regier)
Date: Thu, 18 Jan 1996 11:38:15 -0800
Organization: Qualcomm, Inc.

In article <jinjur-1701962018470001@165.227.52.132>, [e-mail]
(john calhoun) wrote:

> In article <[e-mail]>, [e-mail] (James Shaw) wrote:
>
> > This had an interesting effect on the simulation.  I had to
> > define how elastic collisions worked for a ball striking a surface.
> > I ruled that the spin reverses along the normal, just like translational
> > motion.  It can be anywhere from inelastic to elastic just like
> > translational collisions.  I have this image in my head of a rolling
> > gear catching against the side of something and bouncing back, but
> > otherwise, don't know if this bouncing is physically valid.  Comments?
>
Doesn't sound right to me either.  In an elastic collision, you can't have
nonconservative forces (i.e. friction).  If there are no frictional forces
(really frictional torques), then when the spinning ball collides with the
wall, the ball's angular momentum can't affect the wall.  Therefore,
angular momentum is conserved because no external torques (or angular
impulses) act on your system.

However, in the real world, we do have friction.  When the ball contacts
the surface, a frictional force will be exerted on the ball.  Thus, energy
won't be conserved.  The ball's spin will change because of the frictional
force, yet the  immovable wall won't move because an external impulse acts
to keep the wall in place after the collision.

Since you have an external impulse acting on your system, angular momentum
is not conserved in your collision and the ball's spin will change.  The
extent of change in this collision will depend on the frictional forces
exerted during the collision and the loss in energy as a result of the
collision.

And remember for the linear momentum part, when a ball hits a motionless
wall, you exert a force (really an impulse) to keep the wall in place
during the collision.  So since you're applying an external impulse,
linear momentum is not conserved and that's why your ball bounces back.
But if the collision is elastic, no friction acts on the ball, so energy
is conserved.

> > Rolling had the effect of, with gravity, losing energy in the system.
> > This was at least in my implementation.
>
Well, that's wrong.  Gravity is a conservative force, so energy is
conserved.  If you're losing energy, then that's most likely due to
roundoff error or other inaccuracies in your implementation.  If friction
acts on your ball, as it does on everyday balls when they roll, then your
ball will lose energy as it rolls.  If there's no friction, then your ball
is simply spinning and translating on a surface, and it won't lose its
energy.  You better take a closer look at the simple case of a ball just
bouncing off a surface under the influence of gravity and see whether your
system works right or not.

<whew> I think that's right.

Jason

--
Jason Regier
GlobalStar Software Engineer
QUALCOMM, Inc.
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (John W. Blackburne)
Date: Sun, 21 Jan 1996 17:15:33 +0800
Organization: Tempest

In article <[e-mail]>,
[e-mail] (Jason Regier) wrote:

:In article <jinjur-1701962018470001@165.227.52.132>, [e-mail]
:(john calhoun) wrote:
:
:> In article <[e-mail]>, [e-mail] (James Shaw) wrote:
:>
:> > This had an interesting effect on the simulation.  I had to
:> > define how elastic collisions worked for a ball striking a surface.
:> > I ruled that the spin reverses along the normal, just like translational
:> > motion.  It can be anywhere from inelastic to elastic just like
:> > translational collisions.  I have this image in my head of a rolling
:> > gear catching against the side of something and bouncing back, but
:> > otherwise, don't know if this bouncing is physically valid.  Comments?
:>
:Doesn't sound right to me either.  In an elastic collision, you can't have
:nonconservative forces (i.e. friction).  If there are no frictional forces
:(really frictional torques), then when the spinning ball collides with the
:wall, the ball's angular momentum can't affect the wall.  Therefore,
:angular momentum is conserved because no external torques (or angular
:impulses) act on your system.

I'm not sure I agree with this. I can certainly imagine a ball hitting an
elastic surface at an angle/rotating, so it's surface is initially moving
relative to the elastic surface, and still have it conserving energy, as
long as it doesn't slide along the surface.

Only dynamic friction is non-conservative. Friction between two surfaces
that do not move relative to each other does not lose energy. Enegy may be
lost internally in the rubber (or whatever) as it stretches
perpendicularly and parallel to the surface, and due to other frictional
forces, but it need not be lost in the friction between the ball and
surface. In fact the greater the coefficient of friction the less likely
slipping will be be, so the more likely energy will be conservered.

John
--
John Blackburne,                     [e-mail]
Programmer  Asia, Inc. Online:       http://www.asia-inc.com
Technology consultant and trainer:   http://www.hk.super.net/~johnb

+++++++++++++++++++++++++++

>From [e-mail] (Geoffrey Kinnel)
Date: 25 Jan 1996 21:22:06 GMT
Organization: Drexel University

Bo Lindbergh ([e-mail]) wrote:
: In article <4dk4sc$[e-mail]> [e-mail] (Juri Munkki) writes:
: >
: > As long as you have a limited amount of balls (say under 10), most
: > machines should be able to handle any kind of math you want to throw
: > at them.

: Hmm, what's the greatest number of balls simultaneously in play ever used
: in a physical pinball machine?  And if I ask this in rec.games.pinball,
: will they laugh at me? :-)


: /Bo Lindbergh

As I recall, I heard tell of a machine with a 13-ball multiball! This
sounds suspect to me, but the folks on rec.games.pinball would be happy to
verify or repudiate my claim.  Many machines support three ball multiball
and some support up to six. More than that would get unwieldy, clogging up
the flippers, in my opinion.

Geoff


---------------------------

>From [e-mail] (Richard Drysdall)
Subject: Q: When to call ReleaseResource?
Date: Mon, 22 Jan 1996 16:45:53 +1300
Organization: University of Waikato

Hi.

This may be a silly question, but TR isn't too clear on when to use
ReleaseResource. Suppose I have a function which has a loop in it like
this:


void whatever ()

{
   Handle  resourceHandle;

   do {
      /* UseResFile and some other stuff */
      resourceHandle = Get1Resource ('blah', 128);
      /* look at the resource */
      /* now finished with that resource */
      }
   while (some condition);

} /* end of function */


My question is, should I call ReleaseResource every time I loop? Once at
the end of the function? Or not at all? Is it ok to call ReleaseResource
on a handle, and then 're-create' the resource handle with another call to
GetResource?

Thanks very much for any help.
--
Richard Drysdall, University of Waikato, New Zealand.
"I'm called a theoretical physicist because in practice I'm not."

Information gathering organisations are hereby denied permission to
use any personal information pertaining to myself (including my email
address) in any commercial transaction.

+++++++++++++++++++++++++++

>From Andrew Welch <[e-mail]>
Date: Mon, 22 Jan 1996 00:06:28 -0500
Organization: Ambrosia Software, Inc.

Richard Drysdall wrote:
> =

> My question is, should I call ReleaseResource every time I loop? Once at
> the end of the function? Or not at all? Is it ok to call ReleaseResource
> on a handle, and then 're-create' the resource handle with another call t=
o
> GetResource?

This response is rather lengthy, but hopefully it will be helpful to =

you and others.  It's from my little project "Mac Programming =

Pitfalls".

=2E....

#3 -- Properly handling resources.  Author - Andrew Welch

When you load in a resource -- either directly by calling =

GetResource(), or indirectly by calling a Toolbox routine like =

GetPicture(), which in turn calls GetResource() -- the Resource =

Manager "owns" that Handle.  Do not dispose of it yourself by =

calling DisposeHandle().

The Resource Manager keeps a table of all of the resources that it =

loaded into memory.  If you call GetResource() on a resource Handle =

that is already in memory, the Resource Manager simply returns the =

Handle to you.  When you call CloseResFile(), all of the resource =

Handle=B9s "owned" by that resource file are automatically disposed =

for you.

Therefor, do *not* call DisposeHandle() on any Handle that was =

loaded in by the Resource Manager.  Doing so wil result in the =

Resource Manager referencing a handle that has already been disposed =

of.  Can you say "mysterious random crashes"?

Now let's look at how to properly handle resources in memory.  We'll =

look at some example code first, and then an explanation:

Handle myResource;

myResource =3D GetResource(MY_RESOURCE_TYPE, MY_RESOURCE_ID);
if (myResource !=3D nil)
    {
    HNoPurge(myResource);

// -- Manipulate the resource as you please...

    HPurge(myResource);
    }

The first thing you'll notice is that we don't call ResError() to =

see if the resource was loaded in successfully -- the proper way to =

check to see if a resource was loaded in OK is to check for nil.

Next you'll notice that the first thing we do if the resource could =

be loaded in OK is we make the resource unpurgable.  Any Handle that =

is purgable may disappear out from underneath us if memory is needed =

elsewhere, and we shouldn't assume that that the resource will be =

unpurgable when it is loaded in.

Finally, instead of disposing of the memory that the resource Handle =

occupies, we simply mark it as purgable.  As we already outlined =

earlier, you should not call DisposeHandle() on any resource Handle.  =

If you will never look at the resource again, you can call =

ReleaseResource() to have the Resource Manager dispose of the =

resource Handle in memory, and remove it from the Resource Manager's =

internal data structures.

However a better way to handle it is to simply set the resource =

Handle to purgable as was done in the above example.  What this does =

is it keeps the resource Handle in memory, but if the Memory Manager =

needs more memory for some other operation, it'll automatically =

purge it.

By using this technique, your application will be fast because if =

memory is available, resources that you use will be kept in memory.  =

Calls to GetResource() on handles that are already in memory will =

simply return the resource Handle, rather than go to the disk to =

load the data in.  If for some reason one of your resource Handles =

was purged, subsequent calls to GetResource() will simply load it =

back into memory.

If you use this technique throughout your code, you won't have to =

worry about which Handles to call ReleaseResource() on, which to =

keep around, etc.  It's one simple, effective, optimal technique for =

handling resources throughout your application.  Let the Memory =

Manager and Resource Manager work for you.

+--------------------------------------------------------------+
|    Andrew Welch - Thaumaturgist - Ambrosia Software, Inc.    |
+-------------------------------+------------------------------+
|    AOL-> Keyword: Ambrosia    |  eWorld-> Shortcut: Ambrosia |
|    CIS-> GO word: Ambrosia    |  http://www.AmbrosiaSW.com/  |
+-------------------------------+------------------------------+

+++++++++++++++++++++++++++

>From [e-mail] (David B. Zwiefelhofer)
Date: Tue, 23 Jan 1996 09:04:03 -0500
Organization: Utility Reduction Specialists, Inc.

In article <[e-mail]>,
[e-mail] wrote:

> Hi.
>
> This may be a silly question, but TR isn't too clear on when to use
> ReleaseResource. Suppose I have a function which has a loop in it like
> this:
>
>
> void whatever ()
>
> {
>    Handle  resourceHandle;
>
>    do {
>       /* UseResFile and some other stuff */
>       resourceHandle = Get1Resource ('blah', 128);
>       /* look at the resource */
>       /* now finished with that resource */
>       }
>    while (some condition);
>
> } /* end of function */
>
>
> My question is, should I call ReleaseResource every time I loop? Once at
> the end of the function? Or not at all? Is it ok to call ReleaseResource
> on a handle, and then 're-create' the resource handle with another call to
> GetResource?

Why call Get1Resource in the loop if you're going to need that resource
every time through the loop? Call it once before the loop and
ReleaseResource after the loop.

You could call ReleaseResource after the loop and leave the call to
Get1Resource in the loop. If the resource is in memory already
Get1Resource will give you a handle to it. It won't reload another copy.
So it would load it the first time through the loop and then just keep
giving you a handle to it during the subsequent calls.

Good luck,

Dave

--
David B. Zwiefelhofer
Utility Reduction Specialists, Inc.
1605 Monroe Street, Suite 110
Madison, WI  53211-2052
[phone]
[phone] FAX

+++++++++++++++++++++++++++

>From "Andrew C. Plotkin" <[e-mail]>
Date: Mon, 22 Jan 1996 16:45:53 -0500
Organization: Carnegie Mellon, Pittsburgh, PA

[e-mail] (Stefan Kurth) writes:
> Andrew Welch <[e-mail]> wrote:
>
> > If for some reason one of your resource Handles was purged, subsequent
> > calls to GetResource() will simply load it back into memory.
>
> Will they?  I thought that GetResource would return the empty handle in
> this case, and you would have to call LoadResource explicitly.  No?

>From NIM:

"If the entry contains a a handle whose value is NIL, and if you
haven't called SetResLoad with the load parameter set to FALSE,
GetResource attempts to read the resource into memory."

--Z

"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."

+++++++++++++++++++++++++++

>From [e-mail] (Stefan Kurth)
Date: Mon, 22 Jan 1996 20:45:03 +0100
Organization: none

Andrew Welch <[e-mail]> wrote:

> If for some reason one of your resource Handles was purged, subsequent
> calls to GetResource() will simply load it back into memory.

Will they?  I thought that GetResource would return the empty handle in
this case, and you would have to call LoadResource explicitly.  No?


--
Stefan Kurth
Berlin, Germany

---------------------------

>From Robert Altman <[e-mail]>
Subject: Quickdraw 3D intro??
Date: 23 Jan 1996 20:25:13 GMT
Organization: Internet Online Services

I am an intermediate C programmer, and I would like to write some apps
that use basic Quickdraw 3D (i.e. I would use the standrd viewer). Are
there any good intros to Quickdraw 3D (in bookstores or on the Net)? I
have a book that Apple publishes, but it is difficult to digest as it is
more of a formal definition than a tutorial.

Thanks!!!

--Robert



+++++++++++++++++++++++++++

>From [e-mail] (Eric Jackson)
Date: Wed, 24 Jan 1996 20:52:24 GMT
Organization: NETCOM On-line Communication Services (408 261-4700 guest)

In article <4e3g79$[e-mail]>,
Robert Altman  <[e-mail]> wrote:
>I am an intermediate C programmer, and I would like to write some apps
>that use basic Quickdraw 3D (i.e. I would use the standrd viewer). Are
>there any good intros to Quickdraw 3D (in bookstores or on the Net)? I
>have a book that Apple publishes, but it is difficult to digest as it is
>more of a formal definition than a tutorial.
>
>Thanks!!!

Here is a good place to start.

http://product.info.apple.com/qd3d/QD3D.HTML

Now I would suggest that you take a look at the develop articles
they are pretty good.  I think that you might be able to find them at
the URL that I suggested.

Eric


---------------------------

>From Darren King <[e-mail]>
Subject: Starting MacTCP on startup
Date: Mon, 15 Jan 1996 20:02:29 -0500
Organization: BreakPoint Design

When I start my program, I check for MacTCP.  If it isn't running,
the call returns 0 (correctly).  Most programs start MacTCP if
it isn't running (netscape, etc...).  How do you tell if MacTCP
isn't installed on the system or if just isn't running right
now?  This behaviour is common on home machines that have to dial
up the internet so MacTCP isn't running all the time.

thanx

Darren King, BreakPoint Design

+++++++++++++++++++++++++++

>From [e-mail] (Colen Garoutte-Carson)
Date: Tue, 16 Jan 1996 13:27:56 +0100
Organization: Teleport - Portland's Public Access [phone]

In article <[e-mail]>, [e-mail] wrote:

> When I start my program, I check for MacTCP.  If it isn't running,
> the call returns 0 (correctly).  Most programs start MacTCP if
> it isn't running (netscape, etc...).  How do you tell if MacTCP
> isn't installed on the system or if just isn't running right
> now?  This behaviour is common on home machines that have to dial
> up the internet so MacTCP isn't running all the time.

Darren,

     There is a Gestalt to check for the existance of MacTCP.  However,
the Gestalt will fail under AU/X.  I've include pascal code below for a
hack that I use (to check for the control panel).  This probably won't
work pre-Sys7.

 - Colen Garoutte-Carson


function HasMacTCP: boolean;
   var
      hasit, done: boolean;
      myPB: CInfoPBRec;
      fileName: str255;
      fi: HParamBlockRec;
      result: longint;
   begin
      hasit := Gestalt('mtcp', result) = NoErr;
      if not hasit then    {Failsafe if Gestalt does not work (Which it
does not under AU/X)}
         begin
            fileName := 'Control Panels';
            myPB.ioNameptr := @fileName;
            myPB.ioVRefNum := theWorld.sysVRefNum;
            myPB.ioFVersNum := 0;
            myPB.ioDirID := 0;
            myPB.ioFDirIndex := 0;
            if PBGetCatInfoSync(@myPB) = noErr then
               begin
                  if BitTst(@mypb.ioFlAttrib, 3) then
                     begin
                        fi.ioCompletion := nil;
                        fi.ioNamePtr := @filename;
                        fi.ioVRefNum := theWorld.sysVRefNum;
                        fi.ioDirID := myPB.ioDirID;
                        fi.ioFDirIndex := 1;
                        done := false;
                        while not done do
                           begin
                              done := PBHGetFInfoSync(@fi) <> noErr;
                              if not done then
                                 begin
                                    if (fi.ioFlFndrInfo.fdType = 'cdev')
and ((fi.ioFlFndrInfo.fdCreator = 'atcp') or (fi.ioFlFndrInfo.fdCreator =
'ztcp')) then
                                       begin
                                          hasit := true;
                                          done := true;
                                       end;
                                    fi.ioFDirIndex := fi.ioFDirIndex + 1;
                                    fi.ioDirID := myPB.ioDirID;
                                 end;
                           end;
                     end;
               end;
         end;
      HasMacTCP := hasit;
   end;

+++++++++++++++++++++++++++

>From [e-mail] (Andreas Magnusson)
Date: Wed, 17 Jan 1996 09:10:12 +0100
Organization: -

In article <[e-mail]>,
[e-mail] (Colen Garoutte-Carson) wrote:

>In article <[e-mail]>, [e-mail] wrote:
>
>> When I start my program, I check for MacTCP.  If it isn't running,
>> the call returns 0 (correctly).  Most programs start MacTCP if
>> it isn't running (netscape, etc...).  How do you tell if MacTCP
>> isn't installed on the system or if just isn't running right
>> now?  This behaviour is common on home machines that have to dial
>> up the internet so MacTCP isn't running all the time.
>
>Darren,
>
>     There is a Gestalt to check for the existance of MacTCP.  However,
>the Gestalt will fail under AU/X.  I've include pascal code below for a
>hack that I use (to check for the control panel).  This probably won't
>work pre-Sys7.
>
> - Colen Garoutte-Carson
>
>            fileName := 'Control Panels';

And I doubt it would work on an non-US machine? Or does AU/X require an
US-system so that you know for sure that the folder is called Control
Panels?

Is there some AU/X reason also that you don't use FindFolder instead?

/Andreas

--
Andreas Magnusson         |   ...but Hilta laughed like someone who had
[e-mail]|  thought hard about life and had seen the joke.
[e-mail]       |
[e-mail]       |                               - Terry Pratchett

+++++++++++++++++++++++++++

>From [e-mail] (Stefan Kurth)
Date: Thu, 18 Jan 1996 15:25:58 +0100
Organization: none

Colen Garoutte-Carson <[e-mail]> wrote:

>      Do you know of any safe way, other than Gestalt, to check for the
> existance of MacTCP?

Walk the driver unit table and look for the .ipp driver.


--
Stefan Kurth
Berlin, Germany

+++++++++++++++++++++++++++

>From [e-mail] (Mike Cohen)
Date: Fri, 19 Jan 1996 11:44:46 -0600
Organization: ISIS International

In article <[e-mail]>,
[e-mail] (Andreas Magnusson) wrote:

>And I doubt it would work on an non-US machine? Or does AU/X require an
>US-system so that you know for sure that the folder is called Control
>Panels?

I think A/UX was a restricted product that couldn't be shipped outside the
US.


--
Mike Cohen - [e-mail]
Home Page: http://www.isis-intl.com/
Sound is the same for all the world - Youssou N'dour, "Eyes Open"
*** Save Mystery Science Theater 3000! Comedy Central has decided to
*** cancel it. Call them at [phone] and tell them what you think.
*** Full details at http://fermi.clas.virginia.edu/~jcp9j/canceled.html

+++++++++++++++++++++++++++

>From [e-mail] (Nigel Stanger)
Date: Mon, 22 Jan 1996 13:57:56 +1300
Organization: Dept. of Information Science, University of Otago

In article <[e-mail]>, [e-mail]
(Mike Cohen) wrote:

> I think A/UX was a restricted product that couldn't be shipped outside the
> US.

We had people running it here in New Zealand...

=====================================================================
Nigel Stanger,           Internet: [e-mail]
University of Otago,     Phone: [phone]
Dunedin, NEW ZEALAND.    Fax:   [phone]

+++++++++++++++++++++++++++

>From [e-mail] (Dave Nebinger)
Date: Fri, 19 Jan 1996 14:14:08 -0500
Organization: KHP Services, Inc

In article <[e-mail]>, [e-mail]
(Stefan Kurth) wrote:

> Colen Garoutte-Carson <[e-mail]> wrote:
>
> >      Do you know of any safe way, other than Gestalt, to check for the
> > existance of MacTCP?
>
> Walk the driver unit table and look for the .ipp driver.

Again, this will only determine if MacTCP has been opened already.

It does not determine if it is merely installed but not yet opened.

Dave.

==========================================================
Dave Nebinger                             [e-mail]
             The Alt.Sources.Mac Archivist
     <http://www.AmbrosiaSW.com/alt.sources.mac/>
    <ftp://ftp.AmbrosiaSW.com/pub/alt.sources.mac/>

+++++++++++++++++++++++++++

>From [e-mail] (Colen Garoutte-Carson)
Date: Wed, 17 Jan 1996 15:34:06 +0100
Organization: Teleport - Portland's Public Access [phone]

In article <[e-mail]>,
[e-mail] (Andreas Magnusson) wrote:

> >            fileName := 'Control Panels';
>
> And I doubt it would work on an non-US machine? Or does AU/X require an
> US-system so that you know for sure that the folder is called Control
> Panels?
>
> Is there some AU/X reason also that you don't use FindFolder instead?

Andreas,

     Well, I never made any claims as to the quality of the code.  :)
It's been a while since I threw that together.

     If you want to be technical, all of that failsafe code should be
avoided.  Just because the MacTCP control panel is in the Control Panels
folder doesn't mean it was there when you last booted the machine.

     Do you know of any safe way, other than Gestalt, to check for the
existance of MacTCP?  Since I do not have A/UX, myself, and fixed this
problem for someone from remote, I'm not entirely sure that the Gestalt
does not work.

     Can anyone confirm this for me?

 - Colen

+++++++++++++++++++++++++++

>From [e-mail] (Povl H. Pedersen)
Date: 18 Jan 1996 11:23:54 GMT
Organization: edu!

In article <[e-mail]>,
[e-mail] (Colen Garoutte-Carson) wrote:

:            fileName := 'Control Panels';

nnonononononono NO !!!!!!

ALWAYS use FindFolder to find the special folders (Systemmappe, and those
inside like Kontrolpaneler, Udvidelser, Indstillinger, Skrifter).
Don't ever assume anything about naming of these folders. Or if you do,
then you should also do a:

typedef char struct {
   unused: 1; // we are in the US, so we only do 7-bit ASCII
   data: 7;
}

NeverEver assume anything about such things. Color is misspelled even in the UK.

--
Povl H. Pedersen - [e-mail]
Macintosh / Unix / Win (aaargghh) programmer -  System Administrator
Microsoft philosophy since 1984: Why innovate when you can immitate

+++++++++++++++++++++++++++

>From [e-mail] (Dave Nebinger)
Date: Thu, 18 Jan 1996 09:46:40 -0500
Organization: KHP Services, Inc

In article <[e-mail]>,
[e-mail] (Colen Garoutte-Carson) wrote:

> In article <[e-mail]>, [e-mail] wrote:
>
> > When I start my program, I check for MacTCP.  If it isn't running,
> > the call returns 0 (correctly).  Most programs start MacTCP if
> > it isn't running (netscape, etc...).  How do you tell if MacTCP
> > isn't installed on the system or if just isn't running right
> > now?  This behaviour is common on home machines that have to dial
> > up the internet so MacTCP isn't running all the time.
>
>      There is a Gestalt to check for the existance of MacTCP.  However,
> the Gestalt will fail under AU/X.  I've include pascal code below for a
> hack that I use (to check for the control panel).  This probably won't
> work pre-Sys7.

Note that checking for the control panel does not mean that MacTCP is
installed:

  1.  Inits and cdevs might have been disabled at startup.
  2.  The software might just have been installed without restarting the Mac.

Dave.

==========================================================
Dave Nebinger                             [e-mail]
             The Alt.Sources.Mac Archivist
     <http://www.AmbrosiaSW.com/alt.sources.mac/>
    <ftp://ftp.AmbrosiaSW.com/pub/alt.sources.mac/>

+++++++++++++++++++++++++++

>From [e-mail] (Stefan Kurth)
Date: Tue, 23 Jan 1996 02:13:31 +0100
Organization: none

Dave Nebinger <[e-mail]> wrote:

> [e-mail] (Stefan Kurth) wrote:
>
> > Walk the driver unit table and look for the .ipp driver.
>
> Again, this will only determine if MacTCP has been opened already.
> It does not determine if it is merely installed but not yet opened.

Yes it does.


The MacTCP driver is installed at startup, so its existence or abscence
in the driver table _does_ tell you if MacTCP is installed or not.  To
find out if it has been opened already, you have to check the dOpened
bit of the driver's dCtlFlags.


--
Stefan Kurth
Berlin, Germany

---------------------------

>From [e-mail] (CMSngr)
Subject: VBL Afterlife
Date: 7 Jan 1996 20:29:13 -0500
Organization: America Online, Inc. (1-800-827-6364)


Hi!

     I was wondering what happens to VBL tasks registered from an
application when that application unexpectedly dies. I am seeing some
strangeness in the CW debugger on runs after the first one. Any
comments?

Thanks!
-Chris

+++++++++++++++++++++++++++

>From [e-mail] (Bretton Wade)
Date: Sun, 07 Jan 1996 23:27:22 -0500
Organization: qualia, inc.

In article <4cps19$[e-mail]>, [e-mail] (CMSngr) wrote:

#      I was wondering what happens to VBL tasks registered from an
# application when that application unexpectedly dies. I am seeing some
# strangeness in the CW debugger on runs after the first one. Any
# comments?

Yeah, the VBL sticks around and keeps trying to call the function you gave
it. Patch ExitToShell to remove the task, or reboot after every crash...

--
[e-mail]
http://www.qualia.com/

+++++++++++++++++++++++++++

>From [e-mail] (David Wareing)
Date: Mon, 08 Jan 1996 15:43:59 +0930
Organization: Weyland Yutani

In article <4cps19$[e-mail]>, [e-mail] (CMSngr) wrote:

>Hi!
>
>     I was wondering what happens to VBL tasks registered from an
>application when that application unexpectedly dies. I am seeing some
>strangeness in the CW debugger on runs after the first one. Any
>comments?

What happens is death. Say goodbye to your machine for 3 minutes
while you reboot it from the switch.

This is one of the perils of working with VBL and Time Manager tasks
during development. If you need to 'es' or just get out of the app
in a real hurry, then that TM task is probably going to clobber you
very shortly afterwards. If it doesn't get you immediately, you can
put money on it to crash you within minutes. So, either do without
such tasks during development (substitute them with less accurate
mechanisms), or try to make sure all your debugging code addresses
the tasks and cleans them up properly before exiting.

Then there's still a problem after development too. If a user tries
to do a hard-quit on your app (which seems to be a very common habit
among some users) then it's probably going to be crash time for their Mac.
And of course they'll blame you, despite your protests that hard-quitting
is a stupid thing to do...

--
David Wareing                                 [e-mail]
Belair, South Australia           http://www.AmbrosiaSW.com/~dwareing/
Macintosh Games & Multimedia Programming


+++++++++++++++++++++++++++

>From [e-mail] (Juri Munkki)
Date: 8 Jan 1996 06:26:34 GMT
Organization: Helsinki University of Technology

In article <[e-mail]> [e-mail] (David Wareing) writes:
>In article <4cps19$[e-mail]>, [e-mail] (CMSngr) wrote:
>>     I was wondering what happens to VBL tasks registered from an
>>application when that application unexpectedly dies. I am seeing some
>>strangeness in the CW debugger on runs after the first one. Any
>>comments?

>This is one of the perils of working with VBL and Time Manager tasks
>during development. If you need to 'es' or just get out of the app
>in a real hurry, then that TM task is probably going to clobber you
>very shortly afterwards. If it doesn't get you immediately, you can
>put money on it to crash you within minutes. So, either do without
>such tasks during development (substitute them with less accurate
>mechanisms), or try to make sure all your debugging code addresses
>the tasks and cleans them up properly before exiting.

As was already said, patching ExitToShell works well. Arashi does it for
the sound code, so you can check the source. (It's 68K code only, so
PowerMac programmers will have to do it slightly differently.)

It's much easier to debug a program when you know that doing an ExitToShell
from the debugger is ok.

What I do now is that I have a manager that keeps a list of procedures
to call on ExitToShell. It can also remove calls from the list, so if
the program is terminated normally, the calls are all removed before
ExitToShell is called. All the classes that need this service use this
manager instead of patching ExitToShell on their own. (I would post it,
but I only have a 68K Think C version...)

--
Juri Munkki [e-mail]              Life is easy when polygons are cheap.
http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

+++++++++++++++++++++++++++

>From [e-mail] (Christoph Bergmann)
Date: Mon, 08 Jan 1996 19:02:47 +1000
Organization: University of Queensland

In article <4cqdeq$[e-mail]>, [e-mail] (Juri Munkki) wrote:

> In article <[e-mail]>
[e-mail] (David Wareing) writes:
> >In article <4cps19$[e-mail]>, [e-mail] (CMSngr) wrote:
> >>     I was wondering what happens to VBL tasks registered from an
> >>application when that application unexpectedly dies. I am seeing some
> >>strangeness in the CW debugger on runs after the first one. Any
> >>comments?
>
> >This is one of the perils of working with VBL and Time Manager tasks
> >during development. If you need to 'es' or just get out of the app
> >in a real hurry, then that TM task is probably going to clobber you
> >very shortly afterwards. If it doesn't get you immediately, you can
> >put money on it to crash you within minutes. So, either do without
> >such tasks during development (substitute them with less accurate
> >mechanisms), or try to make sure all your debugging code addresses
> >the tasks and cleans them up properly before exiting.
>
> As was already said, patching ExitToShell works well. Arashi does it for
> the sound code, so you can check the source. (It's 68K code only, so
> PowerMac programmers will have to do it slightly differently.)
>
> It's much easier to debug a program when you know that doing an ExitToShell
> from the debugger is ok.

this would be one alternative..
the one i use was simply to (in the timetask..) check the keyboard. if i
hold down cmd-option-control, the task would kill itself.. saves a lot of
bother with ES trapping, but u wouldnt want it in a final version..

ChrisB

+++++++++++++++++++++++++++

>From [e-mail] (Groovy)
Date: Mon, 08 Jan 1996 16:17:07 -0400
Organization: Groovy Threads

In article <4cqdeq$[e-mail]>, [e-mail] (Juri Munkki) wrote:


>
> What I do now is that I have a manager that keeps a list of procedures
> to call on ExitToShell. It can also remove calls from the list, so if
> the program is terminated normally, the calls are all removed before
> ExitToShell is called. All the classes that need this service use this
> manager instead of patching ExitToShell on their own. (I would post it,
> but I only have a 68K Think C version...)
>
> --
> Juri Munkki [e-mail]              Life is easy when polygons are cheap.
> http://www.iki.fi/~jmunkki                 Windsurfing: Faster than the wind.

My tasks "themselves" always check to see if my app is present and running
before doing anything.   This way if I crash or a user hard quits
my task doesn't care cause it will not do anything but rts.


I guess you could write the code to maybe even dequeue itself.


Groovy

+++++++++++++++++++++++++++

>From [e-mail] (CMSngr)
Date: 10 Jan 1996 20:09:49 -0500
Organization: America Online, Inc. (1-800-827-6364)

[e-mail] says:

"Yeah, the VBL sticks around and keeps trying to call the function you
gave it. Patch ExitToShell to remove the task, or reboot after every
crash..."

Hi!

That was my suspicion. Score one for uninstalling a VBL task after every
sync for debug builds. This may have extra overhead but it sure is safer.

Thanks!
-Chris

PS: What a mac day, first I spent an hour with the Inside Mac
Files book figuring out how to find all the files in a directory and then
this VBL thing. Sometimes the mac is actually worse than embedded
printer code. (My day job.) To bad the elegance of the GUI is not
reflected
in the API eh?

+++++++++++++++++++++++++++

>From [e-mail] (Jesse Jones)
Date: 11 Jan 1996 03:37:21 GMT
Organization: Edmark

In article <[e-mail]>,
[e-mail] (Groovy) wrote:

> In article <4cqdeq$[e-mail]>, [e-mail] (Juri Munkki) wrote:
>
>
> >
> > What I do now is that I have a manager that keeps a list of procedures
> > to call on ExitToShell. It can also remove calls from the list, so if
> > the program is terminated normally, the calls are all removed before
> > ExitToShell is called. All the classes that need this service use this
> > manager instead of patching ExitToShell on their own. (I would post it,
> > but I only have a 68K Think C version...)
> >
> > --
> > Juri Munkki [e-mail]              Life is easy when polygons are
cheap.
> > http://www.iki.fi/~jmunkki                 Windsurfing: Faster than
the wind.
>
> My tasks "themselves" always check to see if my app is present and running
> before doing anything.   This way if I crash or a user hard quits
> my task doesn't care cause it will not do anything but rts.
>
>
> I guess you could write the code to maybe even dequeue itself.
>
>
> Groovy

Note that IM:Processes says that VBL tasks installed in the app's heap are
removed automatically when the app quits or crashes (page 4-9). On the
other hand, the OS doesn't appear to do this for Time Manager tasks: until
I installed an ExitToShell patch I had problems quitting the app
abnormally when Time Manager tasks were installed.

  --Jesse

+++++++++++++++++++++++++++

>From [e-mail] (Bretton Wade)
Date: Mon, 15 Jan 1996 22:46:50 -0500
Organization: qualia, inc.

In article <[e-mail]>,
[e-mail] (Jesse Jones) wrote:

# Note that IM:Processes says that VBL tasks installed in the app's heap are
# removed automatically when the app quits or crashes (page 4-9). On the
# other hand, the OS doesn't appear to do this for Time Manager tasks: until
# I installed an ExitToShell patch I had problems quitting the app
# abnormally when Time Manager tasks were installed.

The normal VBL queue may get cleaned up on exit, but the slot manager
queues don't seem to be.

--
[e-mail]
http://www.qualia.com/

+++++++++++++++++++++++++++

>From Valentin Bonnard <[e-mail]>
Date: 25 Jan 1996 13:08:57 GMT
Organization: Internet Way


[e-mail] (Bretton Wade) wrote:
>In article <[e-mail]>,
>[e-mail] (Jesse Jones) wrote:
>
># Note that IM:Processes says that VBL tasks installed in the app's heap are
># removed automatically when the app quits or crashes (page 4-9). On the
># other hand, the OS doesn't appear to do this for Time Manager tasks: until
># I installed an ExitToShell patch I had problems quitting the app
># abnormally when Time Manager tasks were installed.
>
>The normal VBL queue may get cleaned up on exit, but the slot manager
>queues don't seem to be.
>
>--
>[e-mail]
>http://www.qualia.com/

I have just pressed Cmd-Opt-Esc and pressed the OK button in one of my apps who
install a normal VBL (not with SlotVInstall), and the mouse stoped moving
immediatly... (and I had to do a Cmd-Ctrl-Restart)

Valentin bonnard
[e-mail]


---------------------------

>From [e-mail] (Todd Clements)
Subject: Why would dragging in a thumb crash?
Date: Mon, 22 Jan 1996 21:39:56 -0800
Organization: Harvey Mudd College

I'm trying to work out some of the more hideous bugs in my program right
now, and this one keeps coming up.  I can use the scroll bar just fine in
a certain window...however, if I try to grab the thumb and use it, it
crashes with the resultant macsbug log file:
/* begin log file */
Unimplemented Instruction at 00060258
 Calling chain using A6/R1 links
  Back chain  ISA  Caller
  010B1CF4    68K  00FC26E8  main+00114
  010B1CE0    68K  00FC2734  EventLoop+0000E
  010B1CD0    68K  00FC6AC4  GetEvent+0011C
  010B1CA2    68K  4087C4AA  _FindDialogItem+003C8
  010B1BF2    68K  4086ED2C  _TrackControl+00134
  010B1BAE    68K  4086EB90  _DragControl+00072
  010B1B78    68K  408768DE  _DragTheRgn+001DA
 Closing log
/* end log file */

It doesn't look like its my program thats doing it directly since I'm
letting the system software deal with dragging in the thumb, but I'm sure
I've done it indirectly, somewhere in the depths of memory. Does anyone
have a hint as to what _type_ of error I've made (I realize without code
it would be impossible to tell me the actual error)

Any help is appreciated, as are copies to e-mail.

Todd
--
[e-mail]  Todd Clements - <http://www.cs.hmc.edu/~tclement>
"If the federal government were around when the Creator put His hand on this state, Indiana wouldn't be here...it'd still be waiting for an environmental impact statement. - Ronald Reagan

+++++++++++++++++++++++++++

>From [e-mail] (Mike Duigou)
Date: Wed, 24 Jan 1996 16:35:40 -0800
Organization: FWB Inc.



The usual problem is that you have an action proc when the user is
dragging the thumb. The toolbox is flawed in that you can have an action
proc for the other scroll bar parts, but not the thumb. To track the thumb
you can't do it with an action proc. You *must* specify -1 for the action
proc if the part returned from FindControl is kControlIndicatorPart.
Sample source exists for making the thumb have action when you move it.
(Sorry, I don't have any).

Mike Duigou
FWB Inc.
--
- -
"Me, when I have nothing to say, my lips are sealed." -- Talking Heads

+++++++++++++++++++++++++++

>From [e-mail] (Dave Nebinger)
Date: Thu, 25 Jan 1996 12:52:03 -0500
Organization: KHP Services, Inc

In article <mike_duigou-2401961635400001@205.172.32.172>,
[e-mail] (Mike Duigou) wrote:

> You *must* specify -1 for the action
> proc if the part returned from FindControl is kControlIndicatorPart.

The only flaw in the above statement occurs if you have installed your
action proc into the control.  In that situation you must use
((ControlActionUPP)0).

Dave.

==========================================================
Dave Nebinger                             [e-mail]
             The Alt.Sources.Mac Archivist
     <http://www.AmbrosiaSW.com/alt.sources.mac/>
    <ftp://ftp.AmbrosiaSW.com/pub/alt.sources.mac/>

+++++++++++++++++++++++++++

>From [e-mail] (Adam Winer)
Date: Thu, 25 Jan 1996 14:39:46 -0800
Organization: Oracle Corporation

In article <mike_duigou-2401961635400001@205.172.32.172>,
[e-mail] (Mike Duigou) wrote:

> The usual problem is that you have an action proc when the user is
> dragging the thumb. The toolbox is flawed in that you can have an action
> proc for the other scroll bar parts, but not the thumb. To track the thumb
> you can't do it with an action proc. You *must* specify -1 for the action
> proc if the part returned from FindControl is kControlIndicatorPart.
> Sample source exists for making the thumb have action when you move it.
> (Sorry, I don't have any).

Wrong.  You can have an action proc, but it has a different prototype
than the normal action proc.  Unfortunately, Apple never gave us
support in the headers for this (despite "clearly" documenting it
on page 5-117 of NIM:Toolbox Essentials)

Use the following in your headers:

typedef pascal void (*ControlIndicatorActionProcPtr)(void);

enum {
   uppControlIndicatorActionProcInfo = kPascalStackBased
};

#if GENERATINGCFM
typedef UniversalProcPtr ControlIndicatorActionUPP;

#define CallControlIndicatorActionProc(userRoutine)      \
      CallUniversalProc((UniversalProcPtr)(userRoutine), \
                         uppControlIndicatorActionProcInfo)

#define NewControlIndicatorActionProc(userRoutine)    \
      (ControlIndicatorActionUPP) \
        NewRoutineDescriptor((ProcPtr)(userRoutine), \
                             uppControlIndicatorActionProcInfo, \
                             GetCurrentISA())
#else

typedef ControlIndicatorActionProcPtr ControlIndicatorActionUPP;

#define CallControlIndicatorActionProc(userRoutine)      \
      (*(userRoutine))()
#define NewControlIndicatorActionProc(userRoutine)    \
      (ControlIndicatorActionUPP)(userRoutine)

#endif


-- Adam Winer
[e-mail]

---------------------------

>From [e-mail] (Bretton Wade)
Subject: [ANNOUNCE] BSP Tree Demo Application v1.3
Date: Wed, 24 Jan 1996 23:50:35 -0500
Organization: qualia, inc.

The Macintosh BSP Tree Demonstration application with complete CW 8 source
and projects is available at the URL

file://ftp.qualia.com/pub/bspfaq/BSP_Tree_Demo_1.3.sea.hqx

for more information about BSP Trees, the BSP Tree FAQ is available at the URL

http://www.qualia.com/bspfaq/

--
[e-mail]
http://www.qualia.com/

---------------------------

>From [e-mail] (Julian Harris)
Subject: [Q] PowerPC VBLs?
Date: Wed, 24 Jan 1996 21:19:22 +1300
Organization: Computer Science, The University of Auckland

Hi all,

I've got a bit of code:



// Returns address of a VBL task record, from within a VBL task
pascal long GetVBLRec( void )
    ONEWORDINLINE( 0x2E88 );        // MOVE.L A0, (A7)



That is used inside the VBL routine:


// This is the task that is executed during a VBL
// NOTE: can not have debug information in an interrupt routine!

static pascal void ChangeCursor( void )
    {
    CursorTaskPtr   RecPtr;



    // Get cursor information.
    RecPtr = (CursorTaskPtr) GetVBLRec();


    // Display the next cursor
    SetCursor( *( RecPtr->myCursors[RecPtr->myFrame] ));


    // Advance to the next cursor frame
    RecPtr->myFrame++;


    // Wrap around to the first cursor
    if (RecPtr->myFrame >= RecPtr->numberOfCursors)
        RecPtr->myFrame = 0;


    // Set task to run again
    RecPtr->myVBLTask.vblCount = InterruptDelay;

    }



But I don't know how to do this on the PowerPC. I get a link error on
GetVBLRec which is reasonable because it's 68k ASM.

Anyone have any ideas?

TIA.
.  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .
"I can't remember when I've felt better!"       >  Julian Harris, Programmer   >
- Ralph Wilson after trying new Norecog        >  Comp. Sci. Dept.     x8915  >
  vitamins with special memory inhibitors.    >  The University of Auckland  >
                                             >  [e-mail]    >

+++++++++++++++++++++++++++

>From [e-mail] (Erichsen)
Date: Thu, 25 Jan 1996 13:52:38 -0700
Organization: Coalition to Ban Dihydrogen Monoxide

In article <[e-mail]>,
[e-mail] (Julian Harris) wrote:

>// Returns address of a VBL task record, from within a VBL task
>pascal long GetVBLRec( void )
>    ONEWORDINLINE( 0x2E88 );        // MOVE.L A0, (A7)

You don't need this on a PowerPC. The TaskPtr is passed as a parameter to
your routine on a PowerPC. Like:

pascal  void    ChangeCursor( CursorTaskPtr RecPtr )

---------------------------

>From Kevin <[e-mail]>
Subject: _Jackson Trap
Date: Tue, 16 Jan 1996 13:08:14 -0500
Organization: Interpath -- Providing Internet access to North Carolina

Does anyone have any info on the _Jackson trap ($ABO3).  I had
never heard of it before, but it has been causing some problems
lately and I want to know what the story behind it is.

Please email: [e-mail]

Thanks
kpw

+++++++++++++++++++++++++++

>From [e-mail] (Greg Chapman)
Date: Fri, 19 Jan 1996 11:27:04 -0500
Organization: OHS, Univ. of Western Ontario

In article <[e-mail]>, Kevin <[e-mail]> wrote:

>Does anyone have any info on the _Jackson trap ($ABO3).  I had
>never heard of it before, but it has been causing some problems
>lately and I want to know what the story behind it is.

_Jackson was the code name for Color Quickdraw while it was being
developed. If you're throwing exceptions or bombing out and seeing
_Jackson, you're most likely in a low memory situation, or are doing
something evil to the graf ports et al.

--
Greg Chapman - Mac Programmer/Developer
Systems Manager - U. of Western Ontario OHS
London, Ontario, Canada
email: [e-mail]
- -
"The laws of physics do not apply to this man."

+++++++++++++++++++++++++++

>From [e-mail] (Thurman Gillespy III)
Date: Fri, 19 Jan 1996 10:23:22 -0800
Organization: Dept of Radiology, Univ of Washington

In article <[e-mail]>, Kevin <[e-mail]> wrote:

>Does anyone have any info on the _Jackson trap ($ABO3).  I had
>never heard of it before, but it has been causing some problems
>lately and I want to know what the story behind it is.
>
>Please email: [e-mail]
>
>Thanks
>kpw

Ah, yes, our good friend Mr. _Jackson. Somewhere, somehow (in a galaxy far,
far away) you have caused QuickDraw to crash. Usually because it ran out
of memory. There are canonical methods for handling low memory situations, see
IM Memory for more details. A Q&D workaround is to _always_ leave a
memory cushion whenever you call NewPtr, NewHandle or anything else that
allocates memory. I leave about 50k to be safe, but your milage will most
definitely vary.

--
Thurman Gillespy III             |  [e-mail]
Department of Radiology, SB-05   |  voice [phone]
University of Washington         |  fax   [phone]
Seattle, WA 98195                |

+++++++++++++++++++++++++++

>From [e-mail] (Bill Stewart-Cole)
Date: Sat, 20 Jan 1996 22:43:07 -0600
Organization: ZOG

In article <[e-mail]>, [e-mail]
(Steve Wilson) wrote:

>In article <[e-mail]>, Kevin <[e-mail]> wrote:
>
>> Does anyone have any info on the _Jackson trap ($ABO3).  I had
>> never heard of it before, but it has been causing some problems
>> lately and I want to know what the story behind it is.
>
>I think it is part of color quickdraw.  A lot of crashes occur there, but
>I don't know why that is.

It is not uncommon to crash  CQD by hitting a _Debugger call at the start
of _Jackson when you make a QuickDraw call with the wrong sort of
parameter (like a pointer where a handle should be) or when you are trying
to move large pictures or complex regions around. Deciphering the cause
post-crash is terribly difficult, but you can usually locate it by
sniffing out QuickDraw calls and looking carefully at the last one before
the crash.  What I have most often seen is that CQD is running out of
memory and slamming the stack down into the heap hard (not only thru the
heap end block, but sometimes right down thru the entire heap and out the
bottom end, which gets very ugly) by trying to push an entire picture,
pixmap, or region structure. Because CQD is running from system space it
doesn't run into any problems strictly from overwriting the entire
application heap.

I believe that _Jackson itself is the lowest-level routine for throwing
pixel data into the frame buffer, and some wag at Apple equated this to
Jackson Pollack's spattering of paint on canvas.

--
Bill Stewart-Cole

I dare you to find my web page: Success is its own punishment

+++++++++++++++++++++++++++

>From [e-mail] (Steve Wilson)
Date: 20 Jan 1996 02:42:39 GMT
Organization: Emergent Behavior

In article <[e-mail]>, Kevin <[e-mail]> wrote:

> Does anyone have any info on the _Jackson trap ($ABO3).  I had
> never heard of it before, but it has been causing some problems
> lately and I want to know what the story behind it is.

I think it is part of color quickdraw.  A lot of crashes occur there, but
I don't know why that is.

Steve Wilson
Emergent Behavior
[phone]
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Mike Cohen)
Date: Sat, 20 Jan 1996 14:47:40 -0600
Organization: ISIS International

In article <[e-mail]>,
Kevin <[e-mail]> wrote:

>Does anyone have any info on the _Jackson trap ($ABO3).  I had
>never heard of it before, but it has been causing some problems
>lately and I want to know what the story behind it is.
>
That's an internal call used by Color QuickDraw. When you see the infamous
_Jackson crash, it means QuickDraw either didn't have enough memory to
complete some operation or some software corrupted QuickDraw's globals so
it just decided to give up.


--
Mike Cohen - [e-mail]
Home Page: http://www.isis-intl.com/
Sound is the same for all the world - Youssou N'dour, "Eyes Open"
*** Save Mystery Science Theater 3000! Comedy Central has decided to
*** cancel it. Call them at [phone] and tell them what you think.
*** Full details at http://fermi.clas.virginia.edu/~jcp9j/canceled.html

+++++++++++++++++++++++++++

>From [e-mail] (Marshall Clow)
Date: Sat, 20 Jan 1996 22:30:45 -0800
Organization: Aladdin Systems

In article <[e-mail]>, Kevin <[e-mail]> wrote:

>Does anyone have any info on the _Jackson trap ($ABO3).  I had
>never heard of it before, but it has been causing some problems
>lately and I want to know what the story behind it is.
>
>Please email: [e-mail]
>
Short answer:
   _Jackson is where QuickDraw goes to die.

Slightly longer answer:
   Color QuickDraw calls _Jachson when it is in the middle of doing
something and something goes wrong. An example of this is when it is
drawing a picture, and a region in the picture has to be scaled, and
MapRgn fails (not enough memory, for example).

   Because of some history that I am not privy to, QD jumps to Jackson,
rather than failing gracefully.

   There isn't really anything you can do about it, except make sure that
QuickDraw doesn't fail. 99% of the failures are caused by running out of
memory.
Don't run out of memory.

Not really a satisfying answer, is it?

--
Marshall Clow
Aladdin Systems

"They that can give up essential liberty to obtain a little temporary
 safety deserve neither liberty nor safety." -- Benjamin Franklin
 _Historical Review of Pennsylvania_, 1759

+++++++++++++++++++++++++++

>From [e-mail] (Tom Dowdy)
Date: Tue, 23 Jan 1996 17:03:09 GMT
Organization: Apple Computer, Inc.

In article <[e-mail]>, [e-mail]
(Bill Stewart-Cole) wrote:

> In article <[e-mail]>, [e-mail]
> (Steve Wilson) wrote:
>
> >In article <[e-mail]>, Kevin <[e-mail]> wrote:
> >
> >> Does anyone have any info on the _Jackson trap ($ABO3).  I had
> >> never heard of it before, but it has been causing some problems
> >> lately and I want to know what the story behind it is.
> >
> >I think it is part of color quickdraw.  A lot of crashes occur there, but
> >I don't know why that is.
>
> It is not uncommon to crash  CQD by hitting a _Debugger call at the start
> of _Jackson when you make a QuickDraw call with the wrong sort of
> parameter (like a pointer where a handle should be) or when you are trying
> to move large pictures or complex regions around. Deciphering the cause
> post-crash is terribly difficult, but you can usually locate it by
> sniffing out QuickDraw calls and looking carefully at the last one before
> the crash.  What I have most often seen is that CQD is running out of
> memory and slamming the stack down into the heap hard (not only thru the
> heap end block, but sometimes right down thru the entire heap and out the
> bottom end, which gets very ugly) by trying to push an entire picture,
> pixmap, or region structure. Because CQD is running from system space it
> doesn't run into any problems strictly from overwriting the entire
> application heap.
>
> I believe that _Jackson itself is the lowest-level routine for throwing
> pixel data into the frame buffer, and some wag at Apple equated this to
> Jackson Pollack's spattering of paint on canvas.

Sigh.  I post this every so often, but the incorrect information
still seems to be running around.

No.  _Jackson does *nothing* but throw a SysErr.  At the time
that 32 bit QuickDraw was being written, two traps were assigned to it.
One is now known as _QDDispatch, and the other was never used and kept
its original name _Jackson.  The code name for 32 bit QuickDraw was
Jackson Pollack.  As far as I know, _QDDispatch was never referred
to as _Pollack.  I can only assume that _Jackson was going to be
used for some setjmp style of error recovery, but that was never
implemented.

There are two common ways to end up in _Jackson.  Flat running
out of RAM, or incorrect parameters to QuickDraw.  Because of the
second, you'll sometimes have some bug that is totally unrelated
to drawing cause a _Jackson to fire -- you've trashed either your
heap or your own data, and later attempts to draw same result in
the trap.

--
 Tom Dowdy                  Internet: [e-mail]
 Apple Computer MS:302-3KS  UUCP: {sun,voder,amdahl,decwrl}!apple!dowdy
 1 Infinite Loop            AppleLink: DOWDY1
 Cupertino, CA 95014
 "The 'Ooh-Ah' Bird is so called because it lays square eggs."

---------------------------

>From [e-mail] (Bekiares Tyrone Drew)
Subject: detecting CD-ROM
Date: 18 Jan 1996 22:00:44 GMT
Organization: University of Illinois at Urbana

How would I go about detecting if a machine has a CD-ROM drive
(internal or extenal)?

Should I go through the drive queue and look for removeables
with a certain attribute? I can't think of what would mark
a CD-ROM drive differently than a syquest or bernoulli cartidge
other than read only..


thoughts?

Thanks!
[e-mail]

+++++++++++++++++++++++++++

>From [e-mail] (Marshall Clow)
Date: Sat, 20 Jan 1996 22:38:09 -0800
Organization: Aladdin Systems

In article <4dmfuc$[e-mail]>, [e-mail]
(Bekiares Tyrone Drew) wrote:

>How would I go about detecting if a machine has a CD-ROM drive
>(internal or extenal)?
>
>Should I go through the drive queue and look for removeables
>with a certain attribute? I can't think of what would mark
>a CD-ROM drive differently than a syquest or bernoulli cartidge
>other than read only..
>
Short answer:
   You can't with 100% reliability.

Long answer is at:
   <http://dev.info.apple.com/techqa/devices/dev18.html#RTFToC18>

--
Marshall Clow
Aladdin Systems

"They that can give up essential liberty to obtain a little temporary
 safety deserve neither liberty nor safety." -- Benjamin Franklin
 _Historical Review of Pennsylvania_, 1759

---------------------------

End of C.S.M.P. Digest
**********************