home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
The World of Computer Software
/
The World of Computer Software.iso
/
geob0108.zip
/
NOTE0108.DOC
< prev
next >
Wrap
Text File
|
1993-01-08
|
51KB
|
1,064 lines
----------------------------------------------------------------------
Notes for GoldED 2.41.B0108, BETA release 08 january 1993.
----------------------------------------------------------------------
* THIS IS A BETA RELEASE. A KEY OR BETAKEY IS REQUIRED TO RUN IT !!!
* Welcome to the third and hopefully last beta in the 2.41 beta
series. Unless serious new problems turn up, the public release of
2.41 will be released in the last week of january.
This release mainly contains bugfixes, many of which are very
important fixes to long-standing problems. It has been through a
couple of pre-beta releases in order to weed out possible fatal
"day-after-oops" bugs, and to make sure the Ezycom support works
as it should. My thanks to Jens Mueller for assisting with the
Ezycom pre-beta testing.
!!! DON'T FORGET TO READ ALL NOTES SINCE B1207 !!!
Beta archives and files included in them:
Contained in all archives:
NOTE0108.DOC This document
NEWS0108.DOC Brief new list (short version of this doc)
(Each doc contains all notes for the series)
GEDB0108.ARJ also contains:
GOLDED.EXE GoldED 2.41.B0108, DOS version
GOLDNODE.EXE GoldNODE 2.41.B0108, DOS version
GEOB0108.ARJ also contains:
GED2.EXE GoldED/2 2.41.B0108, OS/2 version
GN2.EXE GoldNODE/2 2.41.B0108, OS/2 version
GED2.ICO Icon for GoldED/2
BGED0108.ARJ also contains:
GOLDED.BDF Updates GoldED B1207 -> B0108 (DOS)
GOLDNODE.BDF Updates GoldNODE B1207 -> B0108 (DOS)
You need BUPDATE.EXE to apply the .BDF's. Unfortunately BUPDATE
is only available in a DOS version.
It was my plan to release a binary update of the OS/2 version
(a BGEO0108.ARJ), but it turned out to that the differences in
the GED2.EXE file were to big for the diff-maker :-(
Archive naming conventions:
BGEDmmdd.* DOS betas with binary difference files.
GEDBmmdd.* DOS betas with complete drop-in exe files.
BGEOmmdd.* OS/2 betas with binary difference files.
GEOBmmdd.* OS/2 betas with complete drop-in exe files.
Have fun and good hunting! :-)
Odinn Sorensen, GoldED author.
----------------------------------------------------------------------
- Fixed a handful of bugs in the internal editor which caused memory
GPF's and memory access violations in the OS/2 version. These bugs
could in the DOS version have been responsible for at least some
of the strange editing behaviour, and in some cases they could
cause corruption of random memory locations, leading to strange
and unpredictable crashes. These bugs have been in GoldED for a
very long time. Thanks to the OS/2 version, they could finally be
located, debugged and fixed - this would have been (and is) next
to impossible in the DOS version.
- Fixed my is_dir() function yet again. It did not recognize the
root (such as C:\) as a valid directory. This most prominently
affected stuff like the SWAPPATH.
- Fixed TZ=ZZZ0 in GoldED dos shells. This was a messy workaround to
a problem concerning some internal time format conversions.
- Fixed Ezycom support to look for the EZYMSG.NOW semaphore and
create/delete it as needed.
- Fixed problem with scanned areas sometimes showing as unscanned in
the arealist.
+ Added more debugging information to the Squish API error logging.
The log now also displays the API source line where the error
occurred. Be sure to send me a cutting from the log (if any was
generated) when reporting errors in the Squish msgbase support.
- Found and fixed a serious bug in my binary search algorithm. This
algorithm is used for many things in GoldED, including all msg
number handling, nodelist lookup, config reader etc. Under certain
circumstances the buggy algorithm would "jump off the rail" and
search way outside the memory locations it was assigned. This
could cause lockups and crashes in DOS and General Protection
Faults (GPF's) in OS/2.
This was a long-standing bug, and it is almost certainly to blame
for many of the reports of lockups and crashes, as well as the
reports of error exits with msg #0 not found and such. If you
still encounter lockups, crashes or GPF's, please report them as
soon as possible and include a step-by-step description of how to
provoke the error to occur. It should be much easier now, because
the bug described above did not have predictable behavior, and
that's gone now.
+ New replacement token, "@os2slash". Is replaced with "/2" in the
OS/2 version and "" (empty string) in the DOS version. Example
usage in a template:
@cfname@os2slash
In my case, this would sign my msgs with "Odinn/2" in OS/2 and
"Odinn" in DOS. Cute, eh? :-)
* GoldED/2: It is currently not possible to use the SCREENSIZE
keyword to switch to 43/50 lines mode. You have to do it manually
with the OS/2 MODE command.
! If GoldNODE runs out of DOS memory, it will try to use XMS and
EMS. If it still cannot get enough XMS/EMS, it would previously
use a disk buffer. The problem is, even a very small disk buffer
can cause dramatic increases in nodelist compile times. Horror
stories of several hours have turned up regularly in the echoes. I
got a bit tired of that...
In this release of GoldNODE, the disk sort buffer is now DISABLED
by default. Instead of trying to use a disk buffer, it will simply
drop the nodes that doesn't fit in the available memory (a beep
and a warning message is given). If for some strange reason you
want to try your luck with the disk sort buffer, a new commandline
option can be used: "-ND" (Enable DISK usage).
! Okay, I give up. The overlay buffer size is now back down to the
original size. If you want better performance, I can recommend to
use the -O commandline option to increase the overlay buffer size,
but check your memory meter and don't let it get below 100k.
----------------------------------------------------------------------
Notes for GoldED 2.41.A1220, PRE-BETA release 20 december 1992.
----------------------------------------------------------------------
* The fixes in this NOTE are relative to the previous PRE-BETA.
- Fixed AREAFILE Ezycom: Origins and AKA's are now set correctly.
- Changed the error logging in the Ezycom support to let me see more
accurately where the access error occurs when saving a msg. I have
not been able to reproduce any access error myself :-(
- Fixed bug when deleting the first msg in Squish areas. There
appears to be a bug in the MSGAPI itself, which breaks some frame
links if the first msg is deleted while the area is locked.
----------------------------------------------------------------------
Notes for GoldED 2.41.A1218, PRE-BETA release 18 december 1992.
----------------------------------------------------------------------
+ GoldED can now display the size of attached files if a msg has the
file attach attribute set. The size (in bytes/kbytes) is displayed
in the bottom header line below each file in the subject line. If
an attached file is missing, GoldED shows "n/a" instead.
+ New keyword: DISPATTACHSIZE <bytes/kbytes/no>". Defaults to
kbytes. Controls how the size of attached files is displayed in
the header. Either the exact byte size, the rounded kbyte size, or
not displayed at all. If the kbytes setting is chosen, the value
is rounded according to the following formula: kbytes = (bytes +
512) / 1024. So a 600 bytes file is rounded up to "1k", but a 500
bytes file is rounded down to "0k".
- The GoldED logo during startup looked bad in codepage 850 and
other codepages that doesn't have single to double line connector
chars. GoldED now checks the current codepage and uses a pure
single-line logo if the codepage does not have single to double
connectors. Currently the following codepages have been determined
to have these connectors: 437, 860, 861, 862, 863, 865. Any other
codepage makes GoldED defaults to single-line.
- Fixed bug in the Squish MSGAPI itself concerning the deletion of
the first msg in Squish areas. I have not been able to reproduce
the reports of this bug, so please let me know if there is still a
problem in this beta.
+ New arealist key command: "AREAjumpnextmatch". When this key is
hit, the selection bar will jump to the next area (if any) that
matches the "fuzzy" quick-search string. This is useful when two
areas that match are wide apart in the arealist.
Suggested key assignment is <Ctrl-Enter> and/or <Tab>:
^Enter AREAjumpnextmatch
Tab AREAjumpnextmatch
Add the above lines to your GOLDKEYS.CFG.
- It was not possible to enter the alpha char using Alt-224 on the
keypad. Fixed.
- Fixed a nasty bug in the internal editor. When undeleting the last
undeletable line, a word (2 bytes) in a random memory location got
zeroed. This, and a related (but non-fatal) bug, resulted in
access violations under OS/2. This one has been in GoldED for a
very long time :-(
- Changed some code concerning the screen blanker / timeout logic.
There have been reports about the blanker and timeout freaking out
when running GoldED past midnight. I have not been able to
reproduce these problems myself, so I don't know if my changes
fixed the problems. Let me know.
- If Windows is detected, the Sound Blaster support in GoldED will
now be disabled.
- A new whitespace triming algorithm was too agressive when saving
your messages. It trimmed everything below ascii '!' off the end
of lines, including such useful characters as '' :-)
- When in the message lister it was once upon a time possible to hit
Alt-L or just L to "redraw" the list and center the selection bar.
Now it is possible again.
- When using a Random System origin such as "ORIGIN @tfname", where
@tfname is the first name of the TO: name, GoldED did not process
the replacement token at the correct time. The tokens in Random
System items were processed immediately after they were read from
the GOLDRAND.CFG file, and that was before the TO: name was
entered.
- Fixed headeredit name/node lookup bugs: Previously acceptable
"name address" form was broken, such as "Some Point .1" (which
would fill in my address field with 2:231/77.1). It also hung when
looking up a blank name field - the default action for this
situation is now to start the node browser at ".?", which is your
current AKA or your Boss's address if you are a point.
- EEEK! Just discovered that my function to detect whether locking
is possible in a particular msgbase path was totally brain-dead.
This resulted in the Hudson msgbase (and lately also the Ezycom
msgbase) never being locked properly before writing to the it.
This could very well have been the cause of _some_ of the reports
of crashed Hudson msgbases. FIXED. This looks like it was a
long-standing bug! :-(
+ GoldED now reports the name of the multitasker it detects, if you
use the -V (verbose) commandline parameter. This is to help you
determine whether your TIMESLICE setting is correct. If you are
running GoldED under a multitasker and have a CPU monitor that
says GoldED is grabbing a lot of CPU time, try this and make sure
your TIMESLICE keyword is set to the name of your multitasker as
described in the manual. If your TIMESLICE setting is correct,
GoldED's CPU use should be very very low when just waiting for
keyboard input.
- Fixed extremely stupid bug concerning the use of AREAS.BBS with
AREAFILE's that need this file to get the echoids in stead of
ECHOMAILxxx. Sorry about this!
- Finally fixed the problem in GoldNODE when using USERLIST's, where
names got reversed and garbled. The source code for that section
has been put back to the way it was in the 2.40 release. Note that
I have not tested this part myself, but only recognized the bug
and fixed it. Please let me know if there are still problems.
+ New replacement token, "@pid". It gives "GED" for the DOS version
and "GED/2" for the OS/2 version. Mostly for use in tearlines.
+ New replacement token, "@longpid". It gives "GoldED" for the DOS
version and "GoldED/2" for the OS/2 version. Mostly for use in
tearlines. The internal default tearline uses "@longpid @version".
----------------------------------------------------------------------
Ezycom specific notes:
! GoldED now only supports the config files of Ezycom v1.02 and
(hopefully) later. The previous beta was only tested with 1.01,
and unfortunately there were a number of significant differences
to 1.02 that I overlooked. This beta is tested to work with 1.02,
and also adds the various watchdog, paging, bad logon etc boards
if defined. If GoldED detects 1.01 or older configs, it will tell
you so and skip it.
- If an Ezycom area was defined but did not exist, GoldED would in
some cases fail to create it and report "another user". Fixed, and
GoldED will now only report "another user" if DOS actually reports
"access denied" when trying to open the msgbase files. Otherwise
it exits and reports the error.
- The EZYCOMMSGPATH and EZYCOMUSERPATH no longer override the paths
in the CONFIG.EZY file, unless they are specified _after_ the
AREAFILE Ezycom.
* Support for the new deleted msgs file (MSGDLTD.BBS) has not yet
been added.
----------------------------------------------------------------------
OS/2 version specific notes:
* Ctrl-Break still terminates GoldED rather brutally. I have tried
trapping it to ensure proper closing of files and such, but I keep
getting access violations. I could also ignore it, but then it
would be impossible to close the process... So be careful with
that key!
- Fixed problem with cursor keys showing alpha chars and generally
not working well. Oddly enough, I never had this problem in my own
OS/2 setup. I suspect different keyboard layouts generate slightly
different key scan codes.
NOTE: Since this problem did not occur in my setup, I cannot be
totally sure it is correctly fixed! Let me know.
- Removed the counter in the statusline. It has not been replaced
with anything else at this time.
- As mentioned in the notes for the previous version, GoldED/2 was a
nasty CPU hog, due to unavoidable keyboard polling and a OS/2
timeslice function that did not work as I expected. This problem
has now been solved to an acceptable degree. Pulse will no longer
hit the roof, but not always hit rock bottom either - the activity
during the keyboard poll between the timeslices can keep it a
couple of percent above zero.
NOTE: To others who may have the same problem with their quickly
ported keyboard polling, timeslicing DOS applications, the
solution is to use DosSleep(1) instead of DosSleep(0) for the
timeslicing. Apparently DosSleep(0) only yields its timeslice to
other processes with the same or higher priority, and Pulse is
naturally at a very low priority. DosSleep(1) creates a 1 ms (in
practice probably more) pause _and_ releases the entire timeslice
to any process. Or so I'm told. But at least it works!
- The intense background colors and border (overscan) color could
not be set. Fixed.
* When GoldED/2 tried to find a GED2.CFG in the previous beta, this
was actually a bug. But now it is a design feature that GoldED/2
looks first for a GED2.CFG and then GOLDED.CFG if the other was
not found.
- Due to a bug in the TopSpeed compiler I use for the OS/2 version,
GoldED could under some circumstances fail to find the correct
GOLDED.CFG file, and sometimes even try to read GED2.EXE or
GOLDED.EXE as a config file :-( Fixed.
- The PID kludge was incorrect for the OS/2 version. It is now GED/2
for the OS/2 version and GED for DOS. The tearline was also not
always set correctly. Now GoldED uses the @longpid (see above)
internally when building the standard tearline.
----------------------------------------------------------------------
----------------------------------------------------------------------
Notes for GoldED 2.41.B1207, BETA release 07 december 1992.
----------------------------------------------------------------------
* This is the second beta in the 2.41 beta series. I'm afraid I got
a bit carried away with the features this time, and did not get as
many bugs fixed as planned. I will try to do better for the next
beta.
Archives and files included in them:
Contained in all archives:
NOTE1207.DOC This document
NEWS1207.DOC Brief new list (short version of this doc)
(Each doc contains all notes for the series)
GEDB1207.ARJ also contains:
GOLDED.EXE GoldED 2.41.B1207, DOS version
GOLDNODE.EXE GoldNODE 2.41.B1207, DOS version
GEOB1207.ARJ also contains:
GED2.EXE GoldED/2 2.41.B1207, OS/2 version
GN2.EXE GoldNODE/2 2.41.B1207, OS/2 version
GED2.ICO Icon for GoldED/2
BGED1207.ARJ also contains:
GOLDED.BDF Updates GoldED B1125 -> B1207 (DOS)
GOLDNODE.BDF Updates GoldNODE P0720 -> B1207 (DOS)
(You need BUPDATE.EXE to apply the .BDF's)
Archive naming conventions:
BGEDmmdd.* DOS betas with binary difference files.
GEDBmmdd.* DOS betas with complete drop-in exe files.
BGEOmmdd.* OS/2 betas with binary difference files.
GEOBmmdd.* OS/2 betas with complete drop-in exe files.
Have fun and good hunting! :-)
* Notes about the OS/2 version (GoldED/2 and GoldNODE/2):
This is the first functional beta release of the OS/2 version of
GoldED. It has not been thoroughly tested, so unexpected behaviour
and bugs will probably be found.
OS/2 IMPLEMENTATION NOTES:
... The Sound Blaster support is currently disabled in the OS/2
version. Later versions may support special SB drivers or
perhaps the MultiMedia extensions. The standard beeping
effects have been ported to the degree it was possible. The
"gwynk" noise when msgs to you are found sounds a bit
different. The BEEPFACTOR is currently not functional.
... Some configuration keywords are not functional in the OS/2
version. I don't have a list of them right now, but it is
mainly keywords associated with hardware.
... Ctrl-Break and Ctrl-C will terminate GoldED/2 immediately,
without warning, and without saving anything before
terminating. Be careful. This will be fixed in later betas.
... If you are using a CPU monitor like Pulse or similar, you will
find that GoldED/2 eats a lot of CPU time (in Pulse the curve
will hit the roof!). I don't know why, because GoldED/2 DOES
release timeslices (using DosSleep(0) calls) in a manner that
works just fine in the DOS versions when running in DOS boxes.
I'm not sure how to fix this annoying problem, but I will do
my best to find out.
... The statusline does not have a memory meter in the OS/2
version. It is currently replaced with a counter that shows
how many times the statusline has been updated. This will
change in later betas, but I'm not sure to what.
... When running in a window on the WPS, the screen update may
seem very sluggish. This is at least partly due to the fact
that CXL (the user interface library I'm using in GoldED)
often writes a character at time, instead of complete lines.
This will probably be optimized a bit in later betas.
... The GoldED/2 icon (GED2.ICO) is designed by myself, but the
G.E.D letters are based on some nice GoldED Windows icons made
by Karsten Palmvig (2:231/78.76). I have taken the liberty of
recoloring them to yellow/gold shades and made them shorter to
fit the OS/2 text under it. There is currently no icon for
GoldNODE/2.
If YOU have some nice icons for GoldED and/or GoldNODE, in
either OS/2 or Windows format, please send them to me or the
nearest reg.site. I will put the GoldED icon collection in the
public release archives.
... GoldED/2 and GoldNODE/2 are compiled with TopSpeed C++ 3.02.
TopSpeed only generates 16-bit applications, but I have
enabled 386 code generation so the speed should be fairly
satifactory.
... GoldNODE/2 has been released in a beta before (in may). This
release contains the fixes (and perhaps bugs) that were
included in the P0720 release.
+ Added support for the Ezycom msgbase! Please note that this is
the first implementation, and there may be some bugs or quirks in
it that I have not found yet. If you find any problem with GoldED
and Ezycom msgbases, please let me know as soon as possible, with
a detailed description of the problem and ways to reproduce it.
EZYCOM MSGBASE IMPLEMENTATION NOTES:
... The Ezycom msgbase support in GoldED is based on structures
and documents for 1.01 and 1.02, but has only been tested by
me with a quick install of Ezycom 1.01.
... Due to the way the Ezycom msgbase is designed, it is not
practically possible to exclude deleted msgs when scanning and
reading msgs. This means that you will always see the deleted
msgs (they will have the DEL attribute set in the header
display and the msg body is replaced with a dummy). To be able
to exclude deleted msgs within the current Ezycom msgbase
design, GoldED would have to read through the entire
MSGHxxxx.BBS file to look at one single DELETED bit in each
msg header. This would take way too long on most systems. I
have discussed this problem with the Ezycom author, and he has
agreed to implement a simple solution for the next Ezycom
release (and beta). I will implement my support for that
solution in the next GoldED beta.
... GoldED does not add to or create the Ezycom userbase. If a
user is not found in the Ezycom USERS.BBS, the user is
assigned the first set of lastreads. No warning is issued
about this. You should add new users by logging on locally
with EZY -L.
... Ezycom userbase lookups have not yet been implemented.
... For manual setup of Ezycom areas, the <msgbase> parameter of
the AREA keyword must be 'E' and the <loc> parameter must be
the board number of the area. For the new AREADEF keyword the
<msgbase> parameter must be 'Ezycom' and <path/board>
paraneter must be the board number. Examples:
AREA EZYCOM "Ezycom support" EE 1
AREADEF EZYCOM "Ezycom support" 0 Echo Ezycom 1 2:231/77 (LOC)
+ New "AREAFILE Ezycom [path to system files]". Reads CONFIG.EZY and
MESSAGES.EZY to get the area setup. Should be compatible with both
1.01 and 1.02, but has only been tested with 1.01.
+ New keyword: "EZYCOMMSGBASEPATH <path>". No default. If not set,
AREAFILE Ezycom will set it. Defines the base path for the Ezycom
msgbase.
+ New keyword: "EZYCOMUSERBASEPATH <path>". No default. If not set,
AREAFILE Ezycom will set it. Defines the base path for the Ezycom
userbase.
* GoldNODE is included in this beta release. Nothing significant was
changed since the P0720 version. If I recall correctly, there were
some bug reports about GoldNODE in P0720. If you find bugs in this
beta version, please let me know in GOLDED.BETA or in netmail, and
I will try to get them fixed. It works fine in my setup (as P0720
have until today).
- Fixed name lookup from the header edit of names containing points,
such as "van.der.somebody". It would work if the name contained a
space, but not without. Note that names in the GoldED nodelist
index is sorted so that "van.der.somebody" goes with the S'es and
not the V's.
- If GoldED finds a DBRIDGE.LRQ file in the AREAPATH, it will try to
find your first USERNAME in it, so that the lastreads can be
updated correctly. In previous versions GoldED would exit with an
error if the name was not found, but it would add you to the file
anyway. Now the error exit has been removed.
- In EDITCOMMENT the following was not possible:
EDITCOMMENT "," "Pause eller grammatisk? ;-)"
It would show and beep the comment constantly. (The comment is in
Danish and refers jokingly to the fact that in Denmark there are
two allowed standards for setting the comma, causing much
confusion).
+ New feature: With 5 new keywords (see below) it is now possible to
specify where and how wide the name/node/date fields in the header
display/edit are. This makes it possible at last to view/edit the
full length (35 chars) of the name field - and this is in fact the
new default in GoldED.
The new keywords and their new defaults are:
DISPHDRNAMESET 8 36 ; old default was 8 28
DISPHDRNODESET 44 16 ; old default was 36 24
DISPHDRDATESET -20 20 ; old default was the same
EDITHDRNAMESET 8 36 ; old default was 8 28
EDITHDRNODESET 44 36 ; old default was 36 44
The syntax of the two parameters of these keywords are:
<keyword> <pos> <len>
Where <pos> is the position on the display line and <len> is the
field width. If <pos> or <len> is a negative value, that value is
_added_ to the display width (such as 80, 100 or 132), so that it
is relative to the right margin. For example a <pos> of -20 on a
80 char display will put it at position 80 - 20 = 60.
NOTE: The attributes display moves along with the DISPHDRNODESET
values.
- Significantly reduced memory requirements for the msg lister. When
run in very_wide*very_tall videomodes like 132x60, it would
usually run out of memory even with the memory meter showing more
than 100k in the statusline...
+ When setting the FRQ attribute in Hudson netmail areas, it is now
again (also) set in the ^aFLAGS kludge. This was removed in P0720
because I found that QuickBBS had defined one of the undefined
attribute bits to FRQ. Unfortunately this new bit was not as
widely supported as ^aFLAGS FRQ.
+ Even if the new FIDONULLFIX keyword is disabled, GoldED will still
fix a NUL if it is the _first_ character in the msg body. This
probably fixes most of these buggy msgs without the overhead of
checking the entire msg.
+ New keyword: "FIDONULLFIX <yes/no>". Defaults to NO. If set to
YES, GoldED will replace NUL chars with LF chars in the msg body
when reading a Fido *.MSG file. This is slower of course, but not
noticably on fast machines.
This option was created to enable GoldED users to read msgs that
were created/handled by brain-dead programs (I don't have names,
sorry) which are not obeying our primary technical standard:
FTS-0001. FTS-1 states that a NUL terminates the msg body. But
sometimes msgs are encountered which have a NUL as the first char
in the msg body or perhaps in other places, thereby causing GoldED
to show a blank or cut-off msg. In some Pascal-based readers, such
as FM, these msgs can be read anyway, because in Pascal a NUL does
not terminate a string like in C/C++.
NOTE: I really ought to make GoldED display a big fat window
complaining when a msg is not conforming to FTS-1 :-(
- Fixed garbage in nodelist lookup when trying to lookup a name or
address which is beyond the end of existing nodes, such as
11111:2/3 or ZZZZZZ.
+ For AREAFILE Squish, it is now possible to specify the filename of
SQUISH.CFG, using a new -C<cfgfile> option. Example:
AREAFILE Squish -cMYSQUISH.CFG
This example will use MYSQUISH.CFG instead of SQUISH.CFG to read
areas.
+ New Random System keyword: "FORCETEMPLATE <yes/no>". Defaults to
NO. This keyword can be used instead of (or together with) the
TEMPLATE keyword in Random System GROUP's. If used, GoldED will
popup the template selection menu when you start writing a new msg
or reply. If you have TEMPLATE's as well as FORCETEMPLATE, you can
use the (random) TEMPLATE's by hitting ESC instead of selecting
from the menu.
NOTE: This keyword ONLY works in GOLDRAND.CFG!
+ New keyword: "AREADESC <setup>". No default. Adds a description
and optionally a group letter, aka, attributes and origin to an
_existing_ area. This is useful if you use an AREAFILE that does
not contain descriptions, groups, akas, attributes or origins.
The full syntax is:
AREADESC <echoid> <"desc"> [group] [aka] [(attrs)] ["origin"]"
<echoid> The echoid.
<"desc"> Area description in quotes.
[group] Optional groupid letter (A-Z), or 0 if not in a
group, or '-' to keep the existing group letter.
[aka] Optional aka address for the area, or '.' for main
aka, or '-' to keep the existing aka.
[(attrs)] Optional default attributes in brackets (), '-' to
keep the existing attributes. Note that the
attributes, if specified, are _added_ to the
default net/echo/local attributes.
["origin"] Optional default origin in quotes.
Note that you cannot skip one of the middle optional parts, even
if you only want to set, say, an origin. Use '-' to keep the
existing value for the parts you skip.
+ New "AREAFILE EchoList [path]<filename>". The [path] defaults to
the AREAPATH. Reads a simple ascii-text file containing an
echolist in this form:
<echoid> <description>
This feature adds descriptions to already existing areas in
GoldED. A common use for this feature might be:
AREAFILE AreasBBS AREAS.BBS
AREAFILE EchoList ECHOLIST.TXT
Descriptions for unknown echoids are ignored. Blank lines and
lines beginning with characters which are illegal in echoids (such
as ';' are also ignored.
+ New commandline parameter: "-W". If "-W" is used, GoldED will
create/overwrite the file GOLDAREA.INC, which will contain all
areas in the new AREADEF form, sorted by your AREALISTSORT
specification. This is very useful for converting your AREAFILE's
to a form you can edit with your favorite text editor and use in
GoldED. It is also useful if you have used the new AREADESC
keyword or the new AREAFILE EchoList reader.
The GOLDAREA.INC file (created in the GOLDPATH) can be used by
adding "INCLUDE GOLDAREA.INC" to your GOLDED.CFG or GOLDAREA.CFG.
When creating the file, GoldED will use '.' if an aka is the same
as the main aka, and leave out the optional origin if it's the
same as the first ORIGIN in your GOLDED.CFG. This makes it easier
to share the same GOLDAREA.INC between different setups.
Try the option now, and look at the output GOLDAREA.INC. It won't
hurt, and will probably be interesting.
+ New commandline parameter: "-M" (for Mute). If "-M" is used,
GoldED will turn off all sounds and noises. Can be turned on again
with "-M-", for example if you have it turned off in your GEDCMD
environment variable and want to turn it on.
- The single '.' in AREADEF's did not work as intended. Fixed.
+ New keyword: "NODELISTWARN <yes/no>". Defaults to YES. If set to
YES, GoldED will warn you during startup if one or more nodelists
are missing. Use NO to disable the warning if it bothers you or
you delete/pack your nodelists when the nodelists are compiled.
Note that GoldED can work fine with lookups etc. without
nodelists as long as it can access its own indexes (GOLDNODE.?IX).
Only the extra details will be missing.
+ Added extra safety measures when shelling to DOS. Now GoldED will
close all msgbase files in _all_ circumstances before shelling,
and re-open afterwards. Previously it closed the msgbase files in
most places, but not all. Some reports indicate that having an
open Hudson msgbase while shelling out, could in some unfortunate
circumstances cause fatal damage to the msgbase if you started
GoldED again while in the GoldED DOS shell :-( This can no longer
happen, because the msgbase files are now 100% _closed_ while in
the DOS shell. The same new safety measures also apply to Squish
and Ezycom msgbases. The Fido msgbase does not need it, because
each *.MSG file is opened, read/written and closed at once.
- If you moved an _unscanned_ msg in a Hudson area, the NETMAIL.BBS
or ECHOMAIL.BBS file would not get updated correctly. This could
cause IMAIL and other scanners to report that the file was
invalid.
- When using the ATTRIBUTES keyword in the GOLDRAND.CFG file, the
attributes would be added to the areas _existing_ attributes (as
set with AREA/AREADEF/AREAFILE. Now they will instead be added to
the default net/echo/local attributes.
- At last! Found and fixed the problem with leading zeroes for
Hudson board numbers in AREAS.BBS:
007 JAMES.BOND 2:231/77.007
The contents of AREAS.BBS can now look like the above without
problems.
- When using more than one COOKIEFILE, only the first one got
correctly indexed. Fixed.
- Fixed bug in AREAFILE D'Bridge for version 1.50.
- Fixed wrong akas in AREAFILE FastEcho for 1.10 and later.
+ New keyword: "COOKIEPATH <path>". Defaults to the GOLDPATH.
Defines the default path for the new COOKIEFILE keyword and the
@random template token. Previously the path for cookies was also
GOLDPATH, but in the @random template token, the default path was
that of the template file. This could give problems if the
templates were not in the GOLDPATH..
+ New keyword: "TEMPLATEPATH <path>". Defaults to the GOLDPATH. Use
this if your templates are not in the GOLDPATH.
- When scanning areas, GoldED would attempt to open a Hudson
msgbase, even if you didn't have any. If it could not open a
Hudson msgbase, it would exit with an error message. No problems
if you already have a Hudson msgbase. Fixed.
! Changed the overlay buffer size again, this time from 150K down to
100K to help operation in low memory situations. If you liked the
improved performance with 150K, just use -O150 on the GoldED
commandline or GEDCMD environment variable. If you still keep
getting out of memory errors, set it to -O64 or lower (down to
about 32).
GENERAL NOTE ABOUT MEMORY:
Watch the memory meter on the statusline (to the left of the
clock). GoldED feels most comfortable if the memory meter shows
more than 100k. However, memory fragmentation can occur if you
read and write msgs for extended periods in one session, and that
could make the memory meter show very low values and still work
okay. This is a very difficult problem to solve, and it will not
be dealt with in the 2.4x series.
----------------------------------------------------------------------
* There were some things I forgot to tell you about in the B1124
notes:
+ Added popup warning if GoldED detects that one or more of your
defined nodelists are missing. (See above for a new keyword to
turn off the warning).
+ Some of the initialization messages previously displayed in the
statusline during startup has been moved up in popup windows. Note
that the related language strings are still named ST_something
instead of IL_something. Just goes to show that the naming of the
language strings could use a complete re-think :-)
! The default overlay buffer size has been increased from about 64K
to 150K. This boosts performance, but may cause out of memory
problems if you are running GoldED in DV windows and such. If that
is the case, use the -O commandline parameter to reduce the
overlay buffer size. (See above for changes and notes).
----------------------------------------------------------------------
----------------------------------------------------------------------
Notes for GoldED 2.41.B1125, BETA release 25 november 1992.
----------------------------------------------------------------------
* Files included in this release:
GOLDED.BDF Binary difference file B1124 -> B1125
BUPDATE.EXE Binary update program
NOTE1125.DOC This doc.
* To apply the update, do the following:
... Place GOLDED.BDF and BUPDATE.EXE in the same directory as
GOLDED.EXE.
... Run BUPDATE GOLDED. If you are running under OS/2, you may
have to add the -I (ignore datetime) option to BUPDATE.
... BUPDATE now fixes your GOLDED.EXE and renames the original to
GOLDED.ORG. You can delete that.
The patch fixes the problems described below:
----------------------------------------------------------------------
- A change in the GOLD*.CFG reader did not allow the use of comments
on lines with keywords. This could result in very strange error
messages and exception errors. Fixed.
----------------------------------------------------------------------
----------------------------------------------------------------------
Notes for GoldED 2.41.B1124, BETA release 24 november 1992.
----------------------------------------------------------------------
* Files included in this BETA release:
GOLDED.EXE GoldED 2.41.B1124 Beta Release
NOTE1124.DOC This doc.
NEWS1124.DOC Brief new list (short version of this doc).
- Fixed Hudson msgbase scanning in the arealist <Alt-S> for both the
QBBSSCAN Old and Rescan options. They gave some frightening
looking results, like zapped lastreads and such. It was not fatal,
but "a bit" shocking to see :-)
+ New keyword: "DISPSTATUSLINE <yes/no>". Defaults to YES. If set to
NO, the statusline with memory meter, clock etc. will be disabled.
This option has been implemented as a temporary help for visually
impaired users. More extensive help will be implemented in later
versions.
+ When GoldED exits with some kind of error, the error description
on the screen was not very informative. Now the description is the
same as the one put in GOLDED.LOG. The logging has been modified
to give a bit more "human readable" meaning, but it is still
primarily designed to give me in your bug reports.
- Fixed the handling of Squish areas with zero msgs. They will now
correctly be reported to contain zero msgs, and entering the area
will no longer give you strange "ghost" users accessing it :-)
+ Tweaked the Squish MSGAPI to report the actual DOS errorcodes, so
that GoldED is able to find out if a real error occurred, or
another user *actually is* accessing the msgbase. Changed the
Squish open area/msg checking to only report "Another user" when
the DOS error code concerns accessing errors, or exit with the
errorcode if not. This should be the final end to the dreaded
"Another User" problem.
- GoldED now makes sure the subject field is correctly terminated
with a NUL if you have a long subject (the max is 71 chars+NUL),
and trailing spaces are now trimmed from the end.
+ Improved the error log description for when the Hudson msgbase
could not be opened or locked. It now tells which directory it is
trying to access the msgbase in, and specifically tells about a
failure to lock it, instead of saying it could not open it.
- Finally got around to fixing the nodelist lookup. Previously the
lookup would almost consistently hit the "wrong" node (one above
or below the obvious choice) if there was no exact match. This was
a long standing bug that got a bit worse in the P0720 update.
- The msg lister didn't show pointnumbers in the header. Fixed.
- Fixed long-standing bug (since 2.40 and probably most 2.32 betas)
in the D'Bridge 1.30 reader in AREAFILE D'Bridge. The .AA1 and
.AA2 files were opened in textmode when they should have been
opened in binary mode. GoldED would therefore in some cases read
garbage or end prematurely and not read all areas.
+ Changed the operation of the TWITMODE keyword, so that the Sound
Blaster feature plays the MsgIsTwit event even if TWITMODE is set
to "Show".
- The file attach dialog window was really messed up. Fixed.
+ New keyword: "AREALISTECHOMAX <size>". Default: 0 (dynamical).
Allows you to specify a fixed width of the EchoID column in the
arealist, or make it dynamically sized (see below). If a negative
value is specified, the width will be the that of the widest
echoid in the arealist plus the negative value. This might be
useful if some long-name echoes have uninteresting ends, such as
"VERYLONGECHOID_R23.PUB" (could benefit from a <size> of -7).
+ The arealist can now dynamically resize the EchoID and Description
columns, so that long echoids are not cut off. The sizing of the
EchoID column is done against the Description column, which
thereby looses or gains width. Previously the EchoID column width
was fixed at 13 chars on 80 column displays, but when we switched
to new and much longer echonames in Denmark recently, this had to
be changed of course ;-) I don't understand why no-one asked for
this feature before!
- When changing the origin at the save menu, the new origin was
inserted with a space missing between the colon and the origin
text, resulting in an invalid origin. Fixed. Argh.. This was the
most visible bug of P0720, because many didn't see it before
saving their msgs :-(
- The origin was truncated at 78 instead of 79 chars. Fixed.
- When using "AREAFILE Squish" and no origin was defined in
SQUISH.CFG, GoldED would put in garbage instead. Fixed. It now
uses the first ORIGIN in GOLDED.CFG instead.
- Spaces in the "word" part of an EDITCOMMENT definition were not
handled correctly. Fixed.
- Fixed long-standing bug in both GoldED and GoldNODE, that took the
point part of an address from the wrong place in some
circumstances, like in SQUISH.CFG: "-p2:231/77 7:77/321.1". This
would set the area aka wrongly to 2:231/77.1. Using .0 would work.
- Fixed bug in handling of the SOUNDDEVICE SB/SBPRO parameters. The
parameters were handled before the driver was loaded, and so they
could never take hold. Also the name of the driver file was stated
incorrectly in the notes - the correct filename is CT-VOICE.DRV.
- Fixed a bug was the very likely cause of at least some of the
strange behaviour of P0720, which I have described as "wild
pointer" problems. It was indeed a wild pointer. It was a silly
bug in the new EVENT keyword handler which put data in regions
outside the properly allocated memory. The EVENT keyword was
introduced in P0720, so this particular "wild pointer" does NOT
account for the strange problems reported for versions earlier
than P0720 (such as leading zeroes in AREAS.BBS not being handled
correctly although the code that does it is 99.99% safe).
+ Speeded up the config reading a bit, especially the parts that add
new areas, such the AREAFILE, AREA and AREADEF keywords.
+ New keyword: "AREADEF <setup>". This is a new version of the
"AREA" keyword, with new and more verbose parameters.
The full syntax is: (must be all on one line of course)
AREADEF <echoid> <"desc"> <group> <type> <msgbase> <path/board>
<aka> <(attrs)> ["origin"]
<echoid> The echoid.
<"desc"> Area description in quotes.
<group> Groupid letter (A-Z), or 0 if not in a group.
<type> Net, Echo or Local.
<msgbase> FIDO, FTSC, HUDSON or SQUISH.
<path/board> Area path, boardnumber or base filename.
<aka> Aka address for the area, or '.' for main aka.
<(attrs)> Default attributes in brackets ().
["origin"] Optional default origin in quotes.
As you can see, it looks a lot like the old AREA keyword. New
stuff is the <group>, the verbose <type> and <msgbase>, the fully
specified <aka> address, the brackets for the attributes and the
optional origin.
Note the possibility of using '.' to specify the main aka. This,
and the optional default origin, makes it simpler to create a
common INCLUDE'able area configuration for several setups with
different addresses, for example two people sharing the same
msgbase.
+ New keyword: "ATTRIBSFRQ <attributes>". No default. Defines the
attributes to use for messages generated with the file request
function. Suggested attributes are: PVT K/S CRA. The FRQ and LOC
attributes are added automatically.
+ The TIMESLICE OS/2 function has been modified so that it now works
even in a booted DOS box (image). A new commandline parameter "-2"
has been added to _force_ the OS/2 timeslice, even if a DOS box is
_not_ detected. Note that using -2 in a regular DOS session (not
under OS/2) may cause GoldED to crash or hang the system. The -2
parameter will be removed in a later version if I can find some
trick to find out whether a DOS session is booted or not. If you
know such a trick, please tell me about it!
The TIMESLICE OS/2 function tests the DOS version to see if it is
running in an OS/2 DOS box. The timeslice function will only be
activated if the reported DOS version is 10.0 or later (20.0 for
OS/2 2.00). The big version number indicates a built-in DOS box
running under OS/2. However a booted DOS image will NOT report
these big versions, only the version of the DOS image (such as
5.00). So unless you have used the SETVER program of MS-DOS 4.0x
or later to tell GoldED it is running under v20.0, you must use
the new -2 parameter as described above.
+ New commandline parameter -FF (Force Force :-). Works like -F
(Force config compile) except that _all_ files are compiled, even
if not out of date. This was added because starting with this
release some files are no longer compiled even with -F unless they
are really out of date.
! To speed up config compiles, GoldED will now only compile
GOLDHELP.CFG if it is really out of date (is newer than
GOLDHELP.GED), or the -FF commandline parameter is used.
- One day I rewrote an "is_dir" function, with the unfortunate and
unexpected result that it didn't work with paths with a backslash
in the end. Fixed. This affected AREAFILE Maximus, the swapper (to
disk) and a few other things.
+ Added support for IMAIL 1.30. Just use AREAFILE IMAIL as usual.
+ Added support for the FMail mail processor (0.90/g and later) by
Folkert J. Wijnstra. Use
AREAFILE FMail [path to FMAIL.CFG and FMAIL.AR] [options]
The FMail support has only been tested with 0.90, but it was built
with structures for 0.92.
! Removed support of the older versions of GEcho in AREAFILE GEcho.
The current GEcho support is based on the structures in the
GEDEV100 archive by the GEcho author. It should however work okay
with any version which uses the SETUP.GE and new AREAFILE.GE
files.
+ New language keyword: "ST_INDEXINGCOOKIE". Defines the string to
display in the statusline when/if GoldED indexes a cookie file.
Defaults to "Indexing Cookie File".
+ New keyword: "COOKIEFILE <[path]<file>>". No default. Path
defaults to the GOLDPATH. Defines any number of "cookie" files,
which will be automatically indexed if needed. The cookie (or
"random") files are those which can be used in the templates,
using the "@random" token.
NOTE: This feature only works for REGISTERED users, and replaces
the external utility "GoldRAND" which was distributed in the
GRAND101.ARJ archive.
! AREAFILE Userdefined has been removed. In my opinion it gave more
confusion and problems than it was worth. There were a couple of
the regular AREAFILE's which used the userdefined stuff
internally, including RemoteAccess and D'Bridge. These now use
real old-style code instead to read the areas. In the case of
RemoteAccess, this should also fix a few bugs..
- When patch-updating GOLDED.EXE using the BUPDATE program, you
could get a garbled screen. This would clear away as soon as you
forced config compiled (-F) or deleted all *.GED files. Fixed.
* GoldED is now compiled with Borland C++ 3.1. Previously it was
compiled with 3.0.
----------------------------------------------------------------------
For the notes since 2.40 up until the first 2.41 beta, see the P0720
note file.
----------------------------------------------------------------------