home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Windows Expert
/
Windows Expert.iso
/
windownt
/
wpj1_7.zip
/
WPJV1N7.TXT
< prev
Wrap
Text File
|
1993-08-02
|
152KB
|
3,829 lines
WW WW WW PPPPPPPP JJ
WW WW WW PP PP JJ
WW WWWW WW PP PP JJ
WW WW WW WW PPPPPPPP JJ
WW WW WW WW PP JJ JJ
WWWW WWWW PP JJ JJ
WW WW PP JJJJJ
----------------------------------------------------------------
The Windows Programmer's Journal Volume 01
Copyright 1993 by Peter J. Davis Number 07
and Mike Wallace Aug 93
----------------------------------------------------------------
A monthly forum for novice-advanced programmers to share ideas
and concepts about programming in the Windows (tm) environment.
Each issue is uploaded to the info systems listed below on the
first of the month, but made available at the convenience of the
sysops, so allow for a couple of days.
You can get in touch with the editors via Internet or Bitnet at:
HJ647C at GWUVM.BITNET or HJ647C at GWUVM.GWU.EDU (Pete)
CompuServe: 71141,2071 (Mike)
GEnie: P.DAVIS5
or you can send paper mail to:
Windows Programmer's Journal
9436 Mirror Pond Dr.
Fairfax, Va. 22032
We can also be reached by phone at: (703) 503-3165.
The WPJ BBS can be reached at: (703) 503-3021.
The WPJ BBS is currently 2400 Baud (8N1). We'll be going to
14,400 in the near future, we hope.
LEGAL STUFF
- Microsoft, MS-DOS, Microsoft Windows, Windows NT, Windows for
Workgroups, Windows for Pen Computing, Win32, and Win32S are
registered trademarks of Microsoft Corporation.
- Turbo Pascal for Windows, Turbo C++ for Windows, and Borland
C++ for Windows are registered trademarks of Borland
International.
- WordPerfect is a registered trademark of WordPerfect
Corporation.
- Other trademarks mentioned herein are the property of their
respective owners.
- WPJ is available from the WINSDK, WINADV and MSWIN32 forums on
CompuServe, and the IBMPC, WINDOWS and BORLAND forums on Genie.
It is also available on America Online in the Programming
library. On Internet, it's available on WSMR-SIMTEL20.ARMY.MIL
and FTP.CICA.INDIANA.EDU. We upload it by the 1st of each month
and it is usually available by the 3rd or 4th, depending on when
the sysops receive it.
- The Windows Programmer's Journal takes no responsibility for
the content of the text within this document. All text is the
property and responsibility of the individual authors. The
Windows Programmer's Journal is solely a vehicle for allowing
articles to be collected and distributed in a common and easy to
share form.
- No part of the Windows Programmer's Journal may be re-published
or duplicated in part or whole, except in the complete and
unmodified form of the Windows Programmer's Journal, without the
express written permission of each individual author. The Windows
Programmer's Journal may not be sold for profit without the
express written permission of the Publishers, Peter Davis and
Michael Wallace, and only then after they have obtained
permission from the individual authors.
Subject Page Author(s)
---------------------------------------------------- ------
WPJ.INI ................................... 4 Pete Davis
Letters ................................... 6 Readers
Beginner's Column ......................... 7 Dave Campbell
Window's Programmer's Guide to Serial...... 17 David Falconer
Pascal in 21 Steps ........................ 22 Bill Lenson
Hacker's Gash ............................. 35 Dennis Chuah
ObjectWindows How-To ...................... 41 Rodney Brown
Creating Help Files ....................... 43 Todd Snoddy
GDI See GDI Do ............................ 48 Bernard Andrys
Windows++ ................................. 57 Philip Sloss
Getting in Touch with Us .................. 62 Pete & Mike
Last Page ................................. 63 Mike Wallace
Windows Programmer's Journal Staff:
Publishers ......................... Pete Davis and Mike Wallace
Editor-in-Chief .................... Pete Davis
Managing Editor .................... Mike Wallace
Contributing Editor ................ Dave Campbell
Contributing Editor ................ Bill Lenson
Contributing Editor ................ Dennis Chuah
Contributing Editor ................ Mike Strock
Contributing Writer ................ Rodney Brown
Contributing Writer ................ Todd Snoddy
Contributing Writer ................ Bernard Andrys
Contributing Writer ................ David Falconer
Contributing Writer ................ Philip Sloss
WPJ.INI
By Pete Davis
Another month, another issue. We didn't do an issue last
month because we didn't have enough articles early enough, but it
gave us a chance to collect a lot of articles for this issue plus
a few for next month's issue. That should give us a good head
start.
This month you'll notice a survey accompanying the magazine.
We're considering putting the magazine in print. Through the help
of a publishing company, this may very well happen. That would
mean some major changes which I want to get out into the open
right away.
First of all, yes, the magazine would cost money. What would
that money get you? It would get you 12 issues a year of a great
magazine. Mike and I would be doing this for a living. That means
that the articles would undergo serious editing, code and text
would be checked and double-checked. We'd have time to make sure
everything is right. At the moment, Mike and I are doing this
part time and can barely afford to spend more than a few hours a
month to work on the magazine.
In our current form, it's almost impossible to do any real
editing. Why? Well, we can't pay the authors, and it's hard
enough to get articles. If we made it difficult, people wouldn't
write for us. Paying the authors would fix that. We'd also be
able to get some of the big names out there, like Andrew
Schulman, Matt Pietrek, etc.
What wouldn't change? One of our hottest areas is help for
beginners. As long as I ever have anything to do with this
magazine, that won't change. We will always be a place where
beginners can turn for information and help. We'd also provide
the more advanced articles that we've been providing over the
past 6 issues.
Going into print would also allow us to reach a much broader
audience than we reach now. We'd be able to get to the people who
can't or haven't found us online.
People who subscribe prior to the first printed issue will
get a discount of about 25% off the regular subscription rate of
around $25. These numbers might change since we're still working
out the details.
We're hoping you will all go for this. In all honesty, I
don't know how long the magazine would be able to survive for
free. Mike and I both do a several things for a living. That
- 4 -
doesn't leave much time to work on the magazine and things are
only getting worse. If we could make our living off the magazine,
we could devote all of our time to it, which we'd both like to
do.
Anyway, fill out the survey and send it to my userid on
CompuServe, the Internet, or via regular mail. Next month we will
draw a name from all the people who responded. The winner will
get a choice between:
- The "Win32 Reference" manuals (a 5-book set)
- A combo of the "Windows API Bible", "Windows Internals", and
"Undocumented Windows"
- The Windows NT Resource Kit
Send the response to me at:
CompuServe: 71644,3570
Internet: 71644.3570@compuserve.com
or regular mail to:
WPJ Survey
9436 Mirror Pond Dr.
Fairfax, Va. 22032 U.S.A.
We look forward to getting your responses.
Peace.
_Pete Davis
- 5 -
Letters
Date: 22-Jun-93 09:10 EDT
From: Scott Guthrie [100060,471]
Subj: Install Program Prob with In use DLL's
Mike and Pete,
I just today downloaded issues 2-5 of your Journal, and I
must say I'm impressed. Unfortunately, you seem to have ditched
the Install program for a variety of reasons, only one of which
I may know the answer to. Microsoft provides an exportable
function in the USER.EXE module of Windows 3.1 called
ExitWindowsExec which will exit Windows and run the executable
indicated in the call and then (when the executable releases
control of the processor) RESTARTS Windows. The executable it
runs can include up to a 128 byte Path and file name, and an
additional 128 bytes of parameters, switches, etc. The
documentation I have read for this function says that it was
included in Win3.1 precisely for the use of SETUP/INSTALL
programs to use when they need to modify code that is currently
in use by Windows (e.g., DLLs). Sorry I wasn't quicker off the
mark on this, but I'm not even sure it is going to help you, as I
haven't been able to get it to work precisely as advertised.
Cheers,
Scott Guthrie
- 6 -
Beginner's Column
By Dave Campbell
Let's see, I know I've got some code here that will show me
how
to put a dialog box into a DLL...no, not that file. Ok, ok, unzip
that entire disk with a -v option into a file, and just search
the thing. Oh, no! Guess I must have dumped it.
I sure hate it when I do that, don't you? Me too. And this
month, since I want to follow up on my promise of putting the
file open dialog box into a DLL, I had to redo it. But that
wasn't all bad, since I'm going to explain it anyway.
The end result of this month's column is going to be code
that runs identical to that of last month, the difference being
that some of the code no longer resides in the executable. To
accomplish this, we are going to pull some of the code from
Hello.EXE and put it into a dynamic link library (DLL) called
FileO.DLL. That's all there is to it. Now for next month...
DLL short-course
DLLs many times are very complex pieces of code, and by
their very nature are reasonably tricky to program. That does not
mean we should be afraid to make use of the power offered to us.
We just need to be careful when using them, and follow the rules.
Most all of us are familiar with ".LIB" files. These are
very much like a collection of ".OBJ" files that we can pull out
during the building of our file, and save having to write the
code again. These LIBs may be locally built, and contain code
personally written, or may be third-party libraries costing
hundreds of dollars. In either case, the code from the LIB
becomes part of your EXE file after the build is completed. If
you use the LIB for three different modules, you will have pieces
of that LIB file in all three of your executables. That takes up
space on your hard disk, and also in memory at execution time.
Execution time? Well, maybe not in DOS, but what if we had
written three TSRs and used a third-party library for some of the
functionality, and then loaded all three? It would be possible to
have, at run-time, three exact copies of the same code in memory.
Kinda wasteful, huh?
I hope I'm not boring everyone, I'm simply trying to set the
stage for the power of a DLL. DLL stands for "Dynamic Link
Library", and as the name implies, the linking takes place
- 7 -
dynamically, or at run-time. In our TSR example above, if DLLs
were available to us in the DOS world, only one copy of the
common code would have been needed, thereby saving disk space
because the EXE files would be smaller, and saving memory at run-
time because the common code would only be loaded once.
That's exactly what we get with DLLs in Microsoft Windows.
When a call is made to a function in a DLL, Windows executes the
function as if the function were local to the window running.
The previous sentence is literally correct, and is one of the
sticky parts of programming DLLs.
DS != SS
I have no intention of digging into this at the depth that
is available in much of the literature, but I am going to try to
give an explanation of what DS != SS means, since it is used
quite often. The primary thing to remember is that DLLs do not
have their own stacks. The calling application's stack gets used
during execution of the DLL. This is a problem because of
near/far references to data. In Windows programming, all near
pointers are assumed to be DS-relative. To avoid a problem with
this, either explicitly declare pointers to be FAR, or cast them
that way as they are used.
Hello.C
First let's take a look at Hello.C. There's been quite a bit
of code removed, and a little added in. I want to discuss the
additions first.
In the include section of Hello.C, a #include of the new
"fileo.h" was added. This is necessary because some definitions
that were in Hello.H are now residing in FileO.H, and our file
needs information about it. Almost any DLL that you will be using
from a third-party vendor will have a DLL included that you will
have to link with. Normally the DLL will have the prototypes of
the functions to call, and any data definitions, or messages that
could be passed.
Next in the global definition area, we define a HANDLE type
named "hFileo". This will be used to identify the DLL after it is
loaded by our program.
HANDLE hFileo;
Next, a prototype was added for "DoFileOpenDlg". This will
be the launch point to get into the DLL to use the dialog box we
- 8 -
created two months ago:
void DoFileOpenDlg(void);
WinMain now contains the following code:
hFileo = LoadLibrary("FILEO.DLL");
if (hFileo < 32)
{
FreeLibrary(hFileo);
MessageBox(GetFocus(),"Dynamic Link Library FILEO.DLL must be
present", szAppName, MB_ICONEXCLAMATION |MB_OK);
return 0;
}
This is the code that loads (possibly) the DLL into memory.
I say possibly because if Windows finds the DLL already loaded,
the currently loaded copy is used. If the load is successful,
LoadLibrary will return a value greater than or equal to 32. This
magic number is HINSTANCE_ERROR, and any return value of this or
lower is an error. On error, we report the reason to the user,
and exit.
Before we exit our WinMain, we must free up the DLL. We are
only concerned with cleaning up after ourselves. Windows will not
allow the DLL to be removed if others are using it. That's what
makes the system so nice to use:
FreeLibrary(hFileo);
WndProc is clean until we get to "case IDM_FILEOPEN":
case IDM_FILEOPEN :
DoFileOpenDlg();
lstrcpy(OutMsg, szFileName);
break;
Instead of calling the dialog box procedure as we have done
in the past, we are now calling a local procedure that will do
that work for us. The code could have been inserted at this
point, but for ease of reading, and comparison, it is cleaner to
have it be isolated.
A change has been made in the reporting of the filename in
that it is transfered locally through the global variable
"szFileName". That will become clear shortly:
DoFileOpenDlg
- 9 -
This procedure is new to Hello, and as I said above, is the
launcher for the DLL function. The mystery takes place in the
first variable declared:
FARPROC lpfnFileOpenDlgProc =
GetProcAddress(hFileo,"FileOpenDlgProc");
This declares a far pointer to a function inside the DLL.
GetProcAddress retrieves that pointer for us by name, given that
we know the handle for the DLL. We know that because we loaded
it, "hFileo". We also know the function name "FileOpenDlgProc",
because we wrote that function two months ago. It just happens to
be external to our program at this point. To execute that
function, we do the following:
(*lpfnFileOpenDlgProc)();
The file name selected will be passed back via Hello.INI,
and we retrieve that into szFileName, to be used above:
GetPrivateProfileString("FileO", "FileName",
"NoName", szFileName, 128, "Hello.ini");
The remainder of Hello.C is identical to the previous code
with the exception of the Fileo.C portions removed.
Hello.DLG
Hello.DLG is now split between Hello.DLG and Fileo.DLG, but
nothing is added or removed, other than the split.
Hello.H
Hello.H is split between Hello.H and Fileo.H.
Hello.DEF
Hello.DEF is identical to before with the exception of the
exported FileOpenDlgProc function being removed.
FILEO
Fileo is a simple DLL and contains the primary pieces that
every DLL contains:
LibMain and WEP
- 10 -
To simplify some of our code, and avoid any nastiness that
may take away from the simplicity, static variables were defined
for the filename and extension, and declared in the beginning of
the Fileo code:
static HANDLE hInst;
static char szFileName[128];
static char szFileSpec[16] = "*.EXE";
static char szDefExt[5] = "EXE";
static WORD wFileAttr = DDL_DRIVES | DDL_DIRECTORY;
LibMain
The LibMain function is called by LibEntry, which is called
by Windows when the DLL is loaded. The LibEntry routine is
provided in the LIBENTRY.OBJ module. LibEntry initializes the
DLL's heap (if a HEAPSIZE value is specified in the DLL's module-
definition file) before calling the LibMain function.
This simply unlocks the data segment of the library (which
is locked by the LocalInit call in LIBENTRY) and returns 1.
int FAR PASCAL LibMain (HANDLE hInstance, WORD wDataSeg,
WORD wHeapSize, LPSTR lpszCmdLine)
{
if (wHeapSize > 0)
UnlockData(0);
hInst = hInstance;
return 1;
}
LibMain is the place to allocate any memory necessary for
use in the DLL, and to do any setup functions.
WEP
WEP is called by Windows when the DLL is released, and
provides a place for us to do cleanup before exiting. In Windows
3.1, WEP is no longer necessary, but if we are trying to write
DLLS to be globally useful, WEP should be included.
int FAR PASCAL WEP (int bSystemExit) {
/*---------------------------------------------------------------
get rid of things you allocated in the LibMain proc here
---------------------------------------------------------------*/
- 11 -
return TRUE;
} /* WEP */
In the case of Fileo, WEP has no functionality.
Fileo Code
The code body for Fileo consists of the two functions
"lstrchr" and "FileOpenDlgProc" from the old Hello.C file.
"lstrchr" has no changes at all, and was simply copied here.
FileOpenDlgProc
This function was stripped straight from the old Hello.C,
modified slightly, and renamed "DoFileOpenDlgProc". It no longer
attempts to open the file selected, the name is simply shipped
back to Hello.C via an ini file, "Hello.INI"
This is all accomplished through the one exported function,
"FileOpenDlgProc"
FileOpenDlgProc
This function is small, and consists of the lines from the
old Hello.C WndProc switch case:
lpfnDoFileOpenDlgProc =
MakeProcInstance(DoFileOpenDlgProc,
hInst);
DialogBox(hInst, "FILEOPEN", GetFocus(), lpfnDoFileOpenDlgProc);
FreeProcInstance(lpfnDoFileOpenDlgProc);
followed by the Hello.INI write of the filename:
WritePrivateProfileString("FileO", "FileName", szFileName,
"Hello.ini");
Fileo.DEF
The DEF file will be recognizable, and has few changes from
the EXE DEF file. As explained above, there is no stack,
therefore no STACKSIZE. WEP must be exported, and our single
exported function "FileOpenDlgProc" is there:
LIBRARY FileO
- 12 -
DESCRIPTION 'FileOpen DLL'
EXETYPE WINDOWS
STUB 'WINSTUB.EXE'
CODE MOVEABLE PRELOAD
DATA MOVEABLE PRELOAD SINGLE
HEAPSIZE 8192
EXPORTS WEP @1 RESIDENTNAME
FileOpenDlgProc
Fileo.mak
The make file is cryptic as ever, but recognizable. The
changes are few, and easy to spot.
Amazingly enough, that's all there is to writing and using a
very simple DLL. I've glossed over many of the nasties, and have
purposefully avoided others. The bottom line is we have a DLL
that's useable and easy to read.
Make Files
I am providing 3 make files with this code. HELLO.MAK and
FILEO.MAK are the Borland make files for Hello.EXE and Fileo.DLL.
MHELLO.MAK is the Microsoft make file for Hello.EXE. Try as I
may, I could not produce a working make file for Microsoft for
FILEO.DLL. The Hello.EXE that I produce with either make file
works with the Fileo.DLL I produce from Borland, but I ran out of
ideas trying to produce a make file for Microsoft for Fileo.DLL.
So, the proof is left to the student. The first one to send me
one that works gets mentioned here next month.
ICONS
Did you really think I was going to leave and not mention
last month's program? Unfortunately, I was so late getting my
article in that I ran the magazine late, and my "Go Suns" hit the
streets after the Bulls took the third in a row. Oh well, the
idea still stands. Animated icons is something we don't see a lot
of in Microsoft Windows. Building all the icon images is probably
the reason why. I was going to take a shot at doing a sweep
second-hand clock, and got bored after 15 seconds worth of
images.
- 13 -
The work for the animated icons starts in WinMain, where a
timer
is started:
SetTimer(hWndMain, ID_TIMER, 500,
MakeProcInstance((FARPROC)TimerProc, hInst));
Windows allows up to 16 timers to be running at any given
time. Normally, SetTimer would be checked for a return value,
and if the value is NULL, display a message about too many
timers.
The first parameter is the handle to the window whose
procedure will receive the WM_TIMER messages.
The second parameter is the timer ID, to identify which of
16 possibles you now have running.
The third parameter is a word specifying the timer interval
in milliseconds (1-65535). In our case, this is 500 msec, or 1/2
second.
The fourth parameter is the instance handle of this window.
InitInstance
Inside InitInstance, we now are loading an Icon, and this is
the "WPJ" icon that will show on the group screen, and for short
periods of time when iconized:
wc.hIcon = LoadIcon(hInstance, "WPJ"); /* Generic Icon
*/
and this value is saved in a global:
hIconMain = wc.hIcon;
WndProc
The WM_PAINT message handler is altered to handle the icon
painting:
case WM_PAINT :
if (IsIconic(hWnd))
{
- 14 -
BeginPaint(hWnd, &ps);
/*---------------------------------------------------------------
Erase the background of the window with what's on the desktop.
This is so the desktop bitmap will show through the transparent
areas of the icon.
---------------------------------------------------------------*/
DefWindowProc(hWnd, WM_ICONERASEBKGND, (WORD)ps.hdc,
0L);
/*---------------------------------------------------------------
Now draw the icon.
---------------------------------------------------------------*/
DrawIcon(ps.hdc, 0,0, hIcon);
EndPaint(hWnd, &ps);
}
Notice that the icon drawn is passed by the generic handle
"hIcon". This will allow us to change the icon image inside the
Timer proc.
One more thing needs to be done inside WndProc, and that is
how to handle the icon when it is dragged on the screen.
Fortunately, Windows sends a message while in that state,
WM_QUERYDRAGICON:
case WM_QUERYDRAGICON:
return((LONG)(WORD)hIconMain);
Consequently, when our WndProc receives that message, we
return the default "WPJ" icon, for the duration of the drag.
TimerProc
The only function our timer proc does in this application is
toggle the two bitmaps GO.ICO and SUNS.ICO:
BOOL FAR PASCAL TimerProc(HWND hWnd, WORD message, WORD wParam,
LONG lParam)
{
static BOOL which = 0;
if (IsIconic(hWnd))
{
if (which = !which)
{
hIcon = LoadIcon(hInst, "GO");
InvalidateRect(hWnd, NULL, FALSE);
- 15 -
SendMessage(hWndMain, WM_PAINT, NULL, NULL);
}
else
{
hIcon = LoadIcon(hInst, "SUNS");
InvalidateRect(hWnd, NULL, FALSE);
SendMessage(hWndMain, WM_PAINT, NULL, NULL);
}
}
}
Depending on the entry value of the static variable "which",
one of the two icons is loaded, and the entire window (the icon
area) is invalidated for painting, and a message is sent to our
own WndProc to execute the WM_PAINT command, which we know from
above will repaint the icon.
Why
All of this is entertaining, sort of, but not useful. I have
simply shown how a timer works, and tried to take some of the
mystery out by doing something stupid.
What if, instead of painting icons on a timer, we were
painting windows? And, what if those windows were over the top of
everything else, so that on a timer, the screen would suddenly
get painted over. Why that sounds like a screen-saver! Well,
kinda-sorta. We wouldn't want to do that on a hard timer, but it
is close. Close enough that, given enough time, there will be the
start of a series in the next issue called "Screen Savers Inside-
Out". One more mystery unfurled.
EndDialog
That's about all there is to say about the code for this
month. Unless something changes, I'm going to discuss common
dialogs and show how to do all this the easy way next time. Feel
free to contact me in any of the ways below. Thanks to those of
you that have e-mailed me questions, keep them coming.
Dave Campbell
WynApse PO Box 86247 Phoenix, AZ 85080-6247 (602)863-0411
wynapse@indirect.com
CIS: 72251,445
Phoenix ACM BBS (602) 970-0474 - WynApse SoftWare forum
- 16 -
Book Review
Title: Windows Programmer's Guide to Serial Communications
Author: Timothy S. Monk
Publisher: SAMS Publishing
By David Falconer
Book Details:
ISBN 0-672-30030-3
Includes Software Disk.
$39.95 USA
I have been searching for ages for a book which would teach
me how to do Comms programming in Windows. Judging by the amount
of questions in the Borland Windows forum on Compuserve, I am not
alone in this quest. The book that seems to be most commonly
recommended is Timothy Monk's "Windows Programmer's Guide to
Serial Communications", so I bought a copy to see what it was
like.
Perhaps I should explain a little about me, so you have an
idea of the role the book was intended to play. I'm not a novice
programmer, but not that experienced either, and I'm very new to
Windows programming. Having worked my way through Petzold, I
decided I was ready to tackle Comms, which is what I wanted to do
in the first place! Also, I learn best by trial and error, and
was looking for something to get me started and that would also
act as a reference later on. I have not started putting the
knowledge I have obtained from this book into practice yet -
priorities change - but I thought I would share what it's like as
a primer.
Firstly, this is not a primer for Windows programming in
general. As a novice, I often found myself with Monk's book in
one hand and Petzold in the other, but the Monk does provide
little snippets of basic stuff that Petzold doesn't cover;
connecting an application to a help file, for example. The other
assumption that is made is that you have some familiarity with
DOS serial Comms - at least the terms used.
The book is laid out in 7 chapters and 3 appendices, over
the course of which you develop a fully working Comms program,
called TSMTERM, starting with the very basics and ending up with
an XMODEM file transfer protocol implementation. Each chapter
starts with an overview of what will be covered, and finishes
with a summary of what has been achieved. There are code
segments everywhere, and these are as much an integral part of
the text as the writing between them is; often it is from the
code that you find exactly how a particular function, or
technique, is implemented. Fortunately all the code, plus fully
compiled executables are provided on the accomanying disk, which
- 17 -
saves on a lot of typing! Each time the code is modified,
advanced upon in the text, a completely new set of source code is
provided on the disk. This means a hefty 3 Megabytes of memory
is needed for the files, but it saves getting confused over which
version of which bit of the program you should be working with at
any one stage. The areas that the chapters cover are:
Chapter 1: Serial Communications
This chapter contains an overview of serial Comms
procedures, starting with the hardware; the RS-232 link, the 3
UARTS commonly found on the PC (8250/16450 and the newer 16550)
in quite a bit of detail and finally the COM ports and their
adresses and interrupts. Monk then goes on to discuss the
principles of DOS Comms programming, with code segments,
including a look at ring buffers. The final part of the chapter
discusses what is going to happen in the rest of the book, by way
of a "design brief" for the TSMTERM program.
Chapter 2: Windows Communications Basics.
This is basically in two major sections. In the first, Monk
outlines the 17 functions of the Windows Communications API,
which are fully documented in Appendix A, and shows how these can
be used in a typical Comms session of open port, send/receive
characters, close port. These initial examples use polling of
the port in the main message loop of the Windows application as
the method of getting information from the port into the
application. Message-based methods, which were introduced with
Windows 3.1, crop up later. In the second half of the chapter,
Monk creates his own API for use with the TSMTERM program, in
order to simplify some of the trickier aspects of using the raw
function calls. These are then all encapsulated in a DLL. This
is done very matter-of-factly, but the code for the LibMain and
WEP functions are provided and briefly explained.
Chapter 3: Terminal Emulation
Again, this chapter is in two sections. The first part
covers terminal emulations, including the full specification of
an ANSI terminal, which is then implemented in a DLL. There is
also a word about finite state machines, on which principles the
ANSI emulation is based. The second part starts assembling the
application from the components created so far. The processing
of the basic Windows messages is covered, as is the Windows
procedure, although this is of quite a complex variety, covering
menu commands as well as the more straight forward messages.
Knowledge of Dialog Box methods are assumed, but I learned quite
a lot about their implementation from studying the code. This is
a fully fledged application, and so includes an "about" box and
- 18 -
interfaces to a basic help file. As well as the ANSI
implementation, the program also allows the user to choose a
basic TTY emulation.
Chapter 4: Errors, Events, and Flow Control
The good practice of calling GetCommError( ) after every
read, write or port poll is introduced and TSMCOMM.DLL, created
in chapter 2, is modified to include error processing. A similar
process is then gone through for communications events using the
GetCommEventMask( ) function and its associates. After both
modifications to the DLL, the main application is also modified
to give the user a choice of how to monitor errors and events.
Finally there is a brief discussion of XON/XOFF and hardware flow
control, after which both methods are added to TSMCOMM.DLL.
Again the main application is modified to give the user a choice
of what to use.
Chapter 5: Message-Based Communications
In Windows 3.1 Microsoft introduced the
EnableCommNotification( ) function which adds a new message to
the Windows system. This is as close as Windows gets to
interrupt driven communications procedures, as the Comms system
is interacted with in response to a received message rather than
having to be polled in the message loop. Chapter 5 gives an
overview of how this system works and then adds the facility to
TSMCOMM.DLL, and modifies the application to give the user the
choice of system.
Chapter 6: Modem Communications
Chapter 6 starts with a look at where modems fit into serial
communications systems. The command set for Hayes modems is then
examined and a modem API is implemented in MODEM.DLL, which
includes setup, dial and hangup routines. A modem commands
section is then added to the main application. As an extention
of this, a dialing directory is implemented, where the user can
store the names of systems, their phone numbers, plus the port
settings: Baud rate, start, stop and data bits. This too, is
then added to the main program.
Chapter 7: File Transfer
To enable the TSMTERM program to download files from the
bulletin boards whose numbers are stored in the newly created
address book, Monk adds the popular XMODEM file transfer
protocol. The chapter starts with a discussion of the protocol;
the packet, error checking, transmission, reception, errors and
timeouts. The protocol is then implemented in another DLL, and
- 19 -
has options for using checksum or CRC error correction. These
functions are then added to the main program. Selection of the
file for transmission is carried out by use of the Common Dialog
Box "Open File", so use of this feature of Windows 3.1 can be
followed by examining the source code.
Appendix A: Windows Serial Communications API Reference
As mentioned above, this provides the documentaion for the
functions and structures provided by Windows for communications
systems. It includes the Windows 3.1 amendments and marks where
the Window 3.1 implementation is different from 3.0.
Appendix B: TSM Communications API Reference.
This provides a full function by function description of
Monk's own API, as used in the TSMTERM program. It describes the
functions created throughout the book which are in TSMCOMM.DLL,
MODEM.DLL and XMODEM.DLL. It is a useful summary and reference
section.
Appendix C: ASCII Character reference.
This is a table in two parts. Part one gives the decimal
value, hex value, display character and description for all of
the 7 bit basic ASCII characters. Part two provides the same
information for the extra members of the extended PC set.
I found the book helpful and easy to follow. The layout is
a bit strange, with funny bits of ornamentation around the edges
of the pages, but you soon get used to it. It goes from
communications basics to quite an advanced program in only 413
pages (including the appendices) so it obviously doesn't hang
around when explaining things to make sure you have got a
particular point. But that's the nice thing about books; if you
find you have missed something, you can always go back and re-
examine it.
As a practical introduction to Windows Communications, I
would heartily reccommend this book. There may be others that do
the job too, but if there are I haven't found them and what I see
here, I like.
I confess that I haven't used the program, TSMTERM, and nor
am I likely to; that's not why I bought the book. I wanted
something that dissected a Windows Comms program, so I would know
how to do it for myself, and that's what I seem to have got. As
I mentioned at the outset, my own project has been "back
burnered" for a while, but I have no qualms about tackling the
Comms aspect head on.
- 20 -
Dialog boxes are another matter though. Ah well, back to
Petzold...
The author can be reached on CompuServe at 100063,1205.
- 21 -
Pascal in 21 Steps
(A Beginner's Column for Turbo Pascal for Windows)
By Bill Lenson
Step Two - Divide and Conquer
REVIEW
Last month we talked about how to write a simple Pascal
program. We showed you what variables and commands are. We also
covered basic data types like integers, characters, and strings.
Even though the examples where quite simple and didn't do
anything practical, the purpose of that column was to introduce
the basic program structure.
This month we pick up the pace quite a bit. By the end of
this column you should be able to read input from the keyboard,
decide if the input is valid, repeat actions over and over, and
break your code up into chunks called functions and procedures.
All this and we'll even show you how to write a simple
calculator.
KEYBOARD IN / SCREEN OUT
Reading information from the keyboard is not that difficult.
There are some tricks to it though to make your life easier.
Last month you used the 'writeln' statement to write text to the
screen. To read text you use 'readln'.
Readln (pronounced 'read line') pauses execution of your
program until you type something and press the <Enter> key. For
example,
Readln(s);
lets you type in some text and when you press <Enter>, puts it
into the variable s. s can be any data type but some are better
than others. If the s 'parameter' to the readln 'procedure' is
an integer then you are only allowed to enter numbers. If it's a
string type you can enter anything. When I say 'only allowed to
enter a number', I mean that you can type in non-numbers but the
program will crash when you press <Enter>. This is bad. Let me
illustrate. Type in the following program.
program Bad_Code;
uses WINCRT;
var
num : integer;
- 22 -
begin
readln(num);
writeln(num);
end.
When you run this program a window will be created and the
cursor will flash. Type in the number 23 and press <Enter>.
Sure enough, 23 gets assigned to the num variable and gets
displayed in the window by the writeln(num) statement. Now,
rerun the program but type in Hello and press <Enter>. CRASH.
Pascal tries to convert Hello into an integer so it can store it
into the num variable. Since it doesn't know how to do this it
bombs. It's a poor program that lets the user crash it so
easily. Try to make the next program blow up. Bet you can't
(poor bet, there is a way but I'm not going to tell you how until
next month).
program Better_Code;
uses WINCRT;
var
numstr : string;
begin
write('Please type a number and press <Enter> : ');
readln(numstr);
write('The number you typed was : ');
writeln(numstr);
end.
No matter what you type in the program spits it back onto
the screen. The only thing limiting in this code is the
programmer has no way to control what the user types. You can't
be sure the user types a number in the above example. We'll show
you how to do that in the next section.
DECISIONS, DECISIONS
Quite often you want a program to do one thing when a
certain input is typed in and do something else when another
thing is entered. The most common way to do this in Pascal is
the IF statement.
Before I show you the IF statement, let me first show you
how you can convert a number stored as a string into an integer
data type. Why we need to do this will become clear shortly.
Try the following code.
program String_2_Number;
uses WINCRT;
var
numstr : string;
- 23 -
num : integer;
errcod : integer;
begin
write('Please type a number and press <Enter> : ');
readln(numstr);
val(numstr, num, errcod);
writeln('The number you typed was : ', num);
{ Notice we combined two write statements into one by
passing }
{ two parameters separated by a comma.
} end.
This program is similar to those above but if you don't
enter a number a zero is printed instead of what you typed in.
The val procedure tries to convert a string into a number, and if
it can't it puts a zero in num and a number greater than zero in
the errcod parameter.
Another thing to note about the above program is the stuff
enclosed in { and }. These are comments and allow the programmer
to describe what is happening in the code. All programs should
have comments that let other programmers (or yourself) look at a
program and tell what the code does. You can put anything in a
comment as long as it starts with a { and ends with a }. As an
aside, sometimes you'll see older Pascal programs with comments
that start with a (* and end in a *). These work too but the
convention these days is to use the braces.
I've been talking for a while about numbers and strings and
how we want a number but we need to enter a string and then
convert it to a number. What's all the hubbub about numbers and
strings? Why not just leave them as strings? The answer is in
the operations you can do to different data types in Pascal. For
instance, you can't add two numbers together if they're stored in
a string. Also, you can't concatenate (stick one string on the
end of another) two numbers together if they're stored in integer
types. Pascal is a typed language meaning that data is always
stored in variables of a certain type. Each type has it's own
set of built-in operations. Some string operations convert a
string to upper case letters from lower case, extract some
characters from the middle of a string, convert a string to a
number, and so on. Most integer functions are math based such as
adding two numbers, subtracting two numbers, multiplying two
numbers, and so on.
This section is about Pascal decision making and program
flow control using the IF statement so let's give a simple
example.
- 24 -
program Better_String_2_Number;
uses WINCRT;
var
numstr : string;
num : integer;
errcod : integer;
begin
write('Please type a number and press <Enter> : ');
readln(numstr);
val(numstr, num, errcod);
if (errcod <> 0) then
writeln('ERROR: The number you entered was invalid.')
else
writeln('The number you typed was : ', num);
end.
The IF statement needs only a little explanation. First the
'<>' in Pascal means 'not equal to'. Also, there are no
semicolon (;) before else's since the whole if..then..else
combination makes a complete statement and last month we said
that semicolons only come at the end of a statement (ugh!). So,
the above is described as, 'If errcod is not equal to zero then
write an error message else write the number entered to the
window'.
In the above, the stuff (I didn't say this was fine
literature) between the IF and the THEN is called the
'condition'. Conditions are logical statements where we compare
things. The things we compare are usually either two variables
or a variable and a value like we do in the above code. Let's
assume we have two variables var1 and var2. Some simple 'logical
operations' we can do are:
Logical Operation Tests if...
================= =====================================
(var1 <> var2) ==> var1 not equal to var2
(var1 = var2) ==> var1 equals var2
(var1 > var2) ==> var1 is greater than var2
(var1 >= var2) ==> var1 is greater than or equal to var2
(var1 < var2) ==> var1 is less than var2
(var1 <= var2) ==> var1 is less than or equal to var2
We'll use logical operations with IF statements many times
in the future so don't worry if they look a little strange right
now. Let's now move on to see how we can repeat something over
and over.
LOOPS
- 25 -
The process of repeating something in Pascal is called
looping. There are four ways to loop but only two of them are
commonly used, one seldom used and one is absolutely forbidden to
be used. The last two will be covered at a later time. This
section presents the first two: the FOR and WHILE loops.
The FOR loop repeats some block of code a fixed number of
times. Let's give a simple example:
program For_loop;
uses WINCRT;
var
x : integer;
begin
for x := 1 to 50 do
write('Hello ');
end.
This program writes 50 'Hello''s on the screen. It's much
simpler than typing out 50 write statements. The FOR statement
is read 'for x equals one to fifty do write hello'. In other
words, x is assigned the number one. The next statement is
executed (the write). x is increased by one and checked to see
if it's greater than 50. If it is, the FOR statement is finished
and the statement after the write is executed. If it isn't, the
write is executed and then x is increased by one and checked to
see if it's greater than 50. And so on. In other words we
repeat the write 50 times while we increase x from 1 to 50. On
the 51st loop, the FOR is exited.
Let's try another.
program For_loop_2;
uses WINCRT;
var
x : integer;
begin
for x := 1 to 10 do
writeln(x);
end.
This example loops 10 times and writes the contents of x
each time through the loop. In other words, the numbers 1 to 10
are written on the screen. The starting number doesn't have to be
1, we could have said:
for x := 5 to 15 do
writeln(x);
which would write the 10 numbers from 5 to 15 on the screen.
- 26 -
The WHILE loop is a little different. WHILE will repeat
until some condition is met. For example:
program While_loop;
uses WINCRT;
var
x : integer;
begin
x := 1;
while (x <= 10) do begin
writeln(x);
x := x + 1;
end;
end.
This program acts just like the For_loop_2 program above.
In it we assign the value one to the x variable. Next, while x
is less than or equal to the value ten (it starts at one so it is
less than 10) repeat everything between the begin and end
statement. What's between the begin and end is the meat of this
WHILE loop. In it we write the x variable to the screen and then
we add one to x. That adding one to x looks a little funny. We
say 'x is assigned the value of x plus one. In other words, we
do the stuff on the right of the ':=' and then put the result in
the variable on the left of the ':='. If we said 'x := x + 2' we
would see every other number from 1 to 10 (i.e., 1, 3, 5, 7, 9).
FOR loops are often used when we know in advance how many
times we will be looping. WHILE loops, on the other hand, are
used when we don't know how many times we'll be looping. In
common programming practice, we would typically use FOR loops to
initialize array variables (a future topic), or repeat something
that never changes in the number of iterations. WHILE loops are
commonly used for reading input from users or files. For
example, reading keyboard input until the user types some exit
command the program knows about. We'll illustrate each looping
technique in more detail next month.
HINDSIGHT is 20-20 in PASCAL
Before we dive into a larger program, I thought I would
point something out about Pascal that beginners must know.
Statements in Pascal cannot see things written after the current
line. On the flip side of the coin, statements can only see
things in the code that occur on lines previous to the current
one. In other words, when the compiler compiles your code into
machine language, it starts at the top and remembers each line as
it comes across it. This is while variables are declared BEFORE
- 27 -
they are used. In the future you'll see that procedures and
functions must be made before they are used. As an example:
var
x : integer;
begin
x := 3;
y := 27;
...
The x:=3 works because x has been declared as an integer
before it was used. The y:=27 causes a compile error because y
has not been declared yet. It may be declared some time later in
the program but that would be too late.
Why I'm bringing this up now is it should help a little bit
when we take a look at procedures and functions. For now, just
declare variables in the var section before you use them in code.
THE CALCULATOR - ATTEMPT #1
Let's take a little break from learning the language to try
to apply what we've learned up until now. The best way to do
that is by writing an example program. From past experience, the
quickest way to learn a language (or a new language) is by
writing example programs. You must type these programs in and
test them for yourselves for maximum learning benefit.
In this section we'll try to build a simple calculator. It
won't do a heck of a lot but I'm sure it will make understanding
of the language easier. OK, so what's this calculator supposed
to do? The first thing is it should ask the user for a number,
an operation to perform, and then another number. Once these
have been entered, the program will spit out the answer. The
operations it can perform are addition (+), subtraction (-),
multiplication (* ), and division (/). One catch to watch out
for is division. You don't want the user dividing by zero as
this is an error and will cause your program to crash. You
therefore need to print an error message yourself if they try to
divide by zero.
Overwhelmed? Don't worry. There are ways to look at
problems and break them down into smaller problems that are more
manageable. Let me show you the program and I'll explain what it
does.
program Calculator_Number_1;
uses WINCRT;
- 28 -
var
numstr1 : string;
numstr2 : string;
operation : string;
num1 : integer;
num2 : integer;
errcod : integer;
begin
writeln('Calculator Example #1');
writeln;
write('Please type a number and press <Enter> : ');
readln(numstr1);
write('Please enter the operation to perform (+, -, *, /) :
');
readln(operation);
write('Please type a number and press <Enter> : ');
readln(numstr2);
val(numstr1, num1, errcod);
val(numstr2, num2, errcod);
if (operation = '+') then
writeln('Answer: ', num1+num2)
else
if (operation = '-') then
writeln('Answer: ', num1-num2)
else
if (operation = '*') then
writeln('Answer: ', num1*num2)
else
if (operation = '/') and (num2 <> 0) then
writeln('Answer: ', num1 div num2)
else
writeln('Error in input. Program ending.');
end.
You should be able to read through this program if you take
it slowly line by line. The first thing we do is declare all
necessary variables. Then we ask the user for each number and an
operation. The user must press <Enter> after each input. Next,
we convert the number strings to integers. Notice we don't check
for an error in converting the number strings, numstr1 and
numstr2. We don't care what they enter. If they enter something
invalid the integer we get from the conversion is just zero.
It's left as an exercise for the reader to put in checks to test
for invalid entries, report the problem and exit the program.
TIP: You might want to look up the HALT procedure if you want to
try this exercise.
OK, after the numbers are converted we perform the series of
IF statement tests to perform the appropriate operation. Check
- 29 -
out the last test for division. The condition statement in that
if has two conditions: (operation = '/') and (num2 <> 0). If we
were to read the last IF statement, it would be described
something like: 'if the operation is division and the num2 isn't
0 then write the answer, else write an error message'. Words
like 'and' which separate conditions will be discussed in more
detail in upcoming months columns.
The programs are starting to look a little messy and I'm not
using any comments (a bad habit but it won't help you to learn if
I always tell you what I'm doing). We need to tidy things up a
bit.
PROCEDURES
A typical Windows program can be several thousand lines
long. If we have one big mainline, things will get beyond messy.
Can you imagine thousands of statements between one begin and
end? I can't. Programmers get around this problem by breaking
up the code into chunks and giving a name to each piece. These
'chunks' are called procedures and functions.
Suppose the mainline of the calculator program above was
written as:
begin
Display_Title;
Get_Inputs;
Convert_Strings_to_Numbers;
Display_Calculations;
end.
Isn't this a little more intuitive? All we need to do is
tell the program what each of these procedures does. This is
simple in this case. We simply take the groups of code that goes
with each procedure and presto.
Here's the next calculator. Notice it's a little bigger but
it's also broken up into nice manageable sections. Also notice,
the procedures are written after the variable declarations but
before the mainline. This way, the procedures can 'see' the
variables because they're declare before it, and the mainline can
'see' the procedures because they're declared before it.
program Calculator_Number_1;
uses WINCRT;
var
numstr1 : string;
- 30 -
numstr2 : string;
operation : string;
num1 : integer;
num2 : integer;
errcod : integer;
procedure Display_Title;
begin
writeln('Calculator Example #1');
writeln;
end;
procedure Get_Inputs;
begin
write('Please type a number and press <Enter> : ');
readln(numstr1);
write('Please enter the operation to perform (+, -, *, /)
: ');
readln(operation);
write('Please type a number and press <Enter> : ');
readln(numstr2);
end;
procedure Convert_Strings_to_Numbers;
begin
val(numstr1, num1, errcod);
val(numstr2, num2, errcod);
end;
procedure Display_Calculations;
begin
if (operation = '+') then
writeln('Answer: ', num1+num2)
else
if (operation = '-') then
writeln('Answer: ', num1-num2)
else
if (operation = '*') then
writeln('Answer: ', num1*num2)
else
if (operation = '/') and (num2 <> 0) then
writeln('Answer: ', num1 div num2)
else
writeln('Error in input. Program
ending.');
end;
begin
Display_Title;
Get_Inputs;
- 31 -
Convert_Strings_to_Numbers;
Display_Calculations;
end.
Voila! Procedures are just like a little program within a
program but they don't get executed until you call them as we do
in the calculator mainline. The mainline is always the first
thing that gets called in a program. It gets called
automatically when the program is started. All other procedures
and functions get called from the mainline. When looking at a
program for the first time, skip down to the mainline and see
what procedures are called. Some may be predefined in the
language such as writeln and others may be user defined as in
Display_Title.
Usually each procedure will have a lot of comments at the
beginning of it. Even though they are often small, it's still
best to make the reader aware of what each procedure does. It's
also good practice to put a ton of comments at the beginning of
the program to describe what it does (a sort of program
overview).
It's interesting to note that a procedure can call another
procedure. Such is the case in Display_Title for instance.
Display_Title calls the procedure writeln. We can write our own
procedures and have another one call it as long as it's declared
BEFORE the one calling it. For instance, we could have called
Display_Title from Get_Inputs but it didn't seem logical to me to
do that. We could have though. This chaining of procedures is a
very powerful feature indeed.
FUNCTIONS
A function is almost exactly the same as a procedure. The
only difference is a function 'returns' a result whereas a
procedure simply gets called and carries out it's instructions.
What do I mean, 'returns a result'? Well, let's look at
math functions such as square root of a number, the sine of an
angle, etc.. Then there's string functions such as get the
string length, concatenate two strings together, etc.. There are
conversion functions such as convert a string to a number,
convert a number to a string, etc.. Functions do something and
then return some result.
Function results have to be of a data type such as integer,
string, char, or any other type. One interesting characteristic
of functions is they can be used in expressions. Expressions are
the right side of assignments (:=) or parameters to functions or
- 32 -
procedures such as writeln. Let's give a couple examples using
the SQRT (square root) function.
x := SQRT(9);
{calculate the square root of 9 and assign that value to x}
x := SQRT(y);
{calculate the square root of the contents of y and assign to
x}
writeln(SQRT(y));
{write the square root of y to the screen}
Some functions can be very simple (such as add the number
three to a variable and return the result), or may be more
complicated (such as a function used to calculate mortgages).
Let's show how to write one of the simple ones this month - a
function that simply returns the math number PI divided by 2. I
think you should be getting the jist of how to write a program
show I'll only show a function and how to call it.
function PI_DIV_2 : integer;
begin
PI_DIV_2 := PI div 2;
end;
...
begin
...
writeln(PI_DIV_2);
...
x := PI_DIV_2 + 4;
...
end.
Instead of procedure, we start a function with the word
function. Also, we must tell what type of value will be
returned. That's the ': integer' part of the header. To tell
the function to return the desired value, we appear to assign a
variable with the same name as the function with the return
result (in this case, the return result is 'PI div 2'). Once the
function is defined, you can use it any way you would a
predefined function.
NEXT MONTH
- 33 -
To wrap up, functions and procedures are very powerful
additions to the Pascal language. We've touched on them here a
little but you will see them over and over in the future and I'm
sure you'll get very used to using them.
Next month I'll show how to pass a parameter to a function
you define, just like SQRT takes one parameter and returns a
result. I'll also show more data types, arrays, constants,
introduce you to controlling the compiler in your code, and
perhaps even how to write a simple game. See you then!
If you wish to contact me, please do so at the following e-
mail addresses:
From Send e-mail to
=========== ===============
Compuserve: >INTERNET:bill.lenson%canrem@uunet.ca
Internet: bill.lenson%canrem@uunet.ca
- 34 -
Hacker's Gash
By Dennis Chuah
This article contains a few of the tips, tricks and traps
from my Windows Programming laboratory book. As I mainly use
Borland C++, most of the discussion here will be slightly biased
to Borland C++. However some issues also concern Windows
programming in general.
It is sad that the Windows API documentation[1] leaves out
certain bits of information that is sometimes crucial to the
correct usage of the API. Take MakeProcInstance for example.
The documentation states this function must only be used to
access functions from instances of the current module. This
implies that MakeProcInstance will create a procedure instance
given a hInstance and an exported procedure address provided it
is called from the task that is associated with the hInstance.
This is not always the case. Take a look at the following code
extract:
Somewhere in the application ...
void CALLBACK SomeProc (void);
.
.
FARPROC lpfnSomeProc;
lpfnSomeProc = MakeProcInstanceForMe (hInstance, (FARPROC)
SomeProc);
if (lpfnSomeProc == SomeProc) MessageBeep (0);
And somewhere else in the application's DLL ...
FARPROC WINAPI MakeProcInstanceForMe (HINSTANCE hInst, FARPROC
lpfnFarProc) {return MakeProcInstance (hInst, lpfnFarProc);
}
MakeProcInstanceForMe will return the address of SomeProc
instead of the procedure instance. This is because
MakeProcInstance will always return the same procedure address
passed to it when it is called from a DLL, regardless of what
hInstance it is passed, a point missed out in the documentation.
Note: the documentation merely states that MakeProcInstance is
not required for library modules. Hence:
Trap: Never use MakeProcInstance to create a procedure instance
from a DLL. Always create the procedure instance first before
passing it to a DLL.
Tip: Use "smart callbacks" prologue and epilogue code if your
compiler supports it. This usually assumes DS is equal to SS.
- 35 -
For most applications this is true, except if you swap data or
stack segments. Callback procedures compiled with "smart
callbacks" do not need procedure instances to bind the data
segment. The compiler just copies SS to DS in the prologue code.
As there is no need for procedure instances, there is no need to
list it in the EXPORTS section of the module definition file.
Tip: Callback procedures in DLLs need not (and should not) use
MakeProcInstance to create procedure instance. The compiler
always assumes that DS is not equal to SS and there is only one
data segment. It will load the correct data segment.
Still on the subject of Windows API, there is a very little
known (and very little documented) header file in Borland C++'s
Include directory called WINDOWSX.H. A wealth of macros can be
found in this header file. For example, the GlobalAllocPtr
macro API allocates memory on the global heap, locks it and
returns a pointer to the locked memory. This is similar to the
malloc function that C programmers are more familiar with. Using
GlobalAllocPtr also saves from having to keep track of a handle.
Simple call GlobalFreePtr when the block of memory is no longer
required. Note: Locking memory has its history in the real mode
version of Windows. In protected mode, locking memory no longer
has any meaning so it is all right to keep a block memory locked
for its lifetime.
Trick: The way GlobalAllocPtr is defined can cause the compiler
to generate a "Code has no effect" warning. To get around this,
place a void typecast over GlobalAllocPtr when you call it.
For example:
DWORD far *lpDword;
.
.
lpDword = GlobalAllocPtr (GHND, sizeof (DWORD) * 20);
.
.
if (lpDword != NULL) (void) GlobalFreePtr (lpDword);
WINDOWSX.H also contains declarations for the Control Macro
API. Using this API instead of sending control messages makes
your code easier to read.
Edit_SetTabStops (hEdit, nTabs, lpTabs);
is more legible than;
SendMessage (hEdit, EM_SETTABSTOPS, (WPARAM) nTabs,
(LPARAM) (const int far *) lpTabs);
- 36 -
In addition, WINDOWSX.H also contains declarations for
message crackers. By using these macros, you will make your code
more portable. I have one criticism though, the macros are not
defined to allow a pointer to be passed to the message handler
functions. If you haven't used WINDOWSX.H before, I strongly
encourage you to start now.
Metafiles:
A metafile is a series of GDI calls. It can be stored as a
physical disk file or represented by a handle to a metafile
(stored in memory).
Tip: Use a drawing package (such as Corel Draw) to generate
metafiles that can be included as a user-defined resource in your
Windows application. This way, you can draw graphics using the
powerful drawing tools of such packages and include the graphic
in your application.
Tip: Including a metafile as a user-defined resource in an
application: Say the metafile name is METAFILE.WMF.
... somewhere in the resource header file (say RES.H):
/* assign an ID value for metafile resource types */
#define METAFILE 2000
#define MYMETAFILEID 100
and somewhere in the resource definition file:
#include "res.h"
.
.
MYMETAFILEID METAFILE "metafile.wmf"
To use the metafile:
#include "res.h"
.
.
HGLOBAL handle;
HMETAFILE hMetafile;
HDC hDc;
RECT rect;
.
.
/* Load the metafile */
handle = LoadResource (hInstance,
FindResource (hInstance, MAKEINTRESOURCE (MYMETAFILEID),
- 37 -
MAKEINTRESOURCE (METAFILE)));
/* no need to retrieve pointer as we only need the handle */
LockResource (handle);
hMetafile = (HMETAFILE) SetMetafileBitsBetter ((HMETAFILE)
handle);
UnlockResource (handle);
/* Then draw it */
hDc = GetDC (hWnd);
GetClientRect (hWnd, &rect);
SetMapMode (hDc, MM_ANISOTROPIC);
SetViewportExt (hDc, rect.right, rect.bottom);
PlayMetaFile (hDc, hMetafile);
ReleaseDC (hWnd, hDc);
/* Clean up */
DeleteObject (hMetafile);
FreeResource (handle);
It is important to set the mapping mode to MM_ANISOTROPIC
and the viewport extents to rectangle in which the metafile is to
be drawn. This ensures that the whole metafile is drawn inside
the specified rectangle. However, some metafiles behaves badly
and will draw outside the specified rectangle. Most metafiles
sets the window extent and origin, and draws within the limits of
the coordinates specified. However, this behaviour cannot be
guaranteed for all metafiles.
Trap: Forgetting to call LockResource can cause
SetMetafileBitsBetter to return NULL.
Trap: Metafiles with placeable headers cannot be processed by
Windows so make sure you save your metafile without the placeable
header.
Miscellaneous:
Trap: Be very careful when allocating memory in a DLL. Memory
allocated by a DLL belongs to the application that called the
DLL. Passing the handle or pointer to the block of memory to
another application will most likely cause a GP fault. If a DLL
needs to share memory between applications, use the GMEM_SHARE
flag when calling GlobalAlloc. This, however, poses another
problem. Memory blocks allocated with the GMEM_SHARE flag,
although can be shared, still belongs to the application that
allocated it and will be freed when the application terminates.
If another application calls the DLL to access the memory after
the application that allocated the memory terminates, a GP fault
- 38 -
will occur. The following diagram better illustrates this (see
the Help file - Ed.).
Tip: So, the lessons here are:
a) if possible, let the application allocate memory and pass the
handle/pointer to the DLL
b) if the DLL must allocate memory, make sure it is used only for
the application that caused it to be allocated
c) if the DLL has to allocate memory that will be shared between
applications, write a small application that allocates memory on
behalf of the DLL. This application has to be active for at
least the life of the DLL. (See accompanying article --
Allocating Shareable Memory (in next month's issue - Ed.))
Notes:
[1] - I define Windows API documentation to be information
documented in Windows API Vols 1-3. I know there are other
sources of information, such as the numerous articles from
Microsoft. However, it is my opinion that any information that
is not included in a function's entry in the Windows API manuals
should be classified as "undocumented". If the set of manuals
are to be the primary source of reference material for Windows
programming, it would be safe to assume that information
documented elsewhere is not widely accessible to Windows
programmers and therefore undocumented.
About the author:
I am currently doing a PhD. degree in Electrical And Electronic
Engineering at the University Of Canterbury. My programming
experience dates back to the days when "real programmers" coded
with Z80 machine code. For the past two years I have taken an
interest in Windows programming. Please send any comments,
questions, criticisms to me via:
email: chuah@elec.canterbury.ac.nz
or post mail:
Dennis Chuah
c/o The Electronic and Electrical Engineering Department
University of Canterbury
Private Bag
Christchurch
New Zealand.
- 39 -
All mail (including hate mail) is appreciated. I will try
to answer all questions personally, or if the answer has general
appeal, in the next issue. If you are sending me e-mail, please
make sure you provide me with an address that I can reach (most
Internet and Bitnet nodes are reachable, and so is CompuServe).
- 40 -
Book Review
Title: ObjectWindows How-To
Author: Gary Syck
Publisher: Waite Group Press
By Rodney M. Brown
The other day, I was in my local WaldenBooks bookstore,
browsing over the crop of C/C++ books. One book that caught my
attention was "ObjectWindows How-To". This book offers
hints/tips on programming in Borland C++/Turbo C++ with
ObjectWindows.
The book is organized by problems, with the solution or
"How-To" following. Each solution is graded in complexity; Easy,
Moderate and Hard. After the solution is explained, it is
followed with the author's C++ code. The C++ code is great for
beginners. The author placed comments throughout the code
explaining what each function does. The author also wrote a 2-3
page explanation of the code he wrote.
This book is a great reference. It is filled with solutions
to printing, changing the appearance/color of windows, etc. Below
is a list of all the problems that are explained in the book:
1. Making An ObjectWindows Application.
2. Changing the Class Information & Style of a Window.
3. Adding control windows to a window.
4. Adding menus to a program.
5. Creating an MDI application.
6. Making and Using DLLs.
7. Using the TDialog object class.
8. Making a file selection dialog.
9. Making a text search dialog.
10. Using text controls.
11. Using list box and combo box controls.
12. Using check box and radio button controls.
13. Using string tables.
14. Converting programs to another language.
15. Loading different menus from a resource file.
16. Adding and Deleting menu items.
17. Changing the attributes of a menu item.
18. Changing the check marks in menus.
19. Using bitmaps as menu items.
20. Creating pop-up menus.
21. Creating a text editor.
22. Getting input from the keyboard.
23. Getting input from the mouse.
24. Using the serial port.
25. Using the sound interface.
26. Handling the printer.
- 41 -
27. Drawing in a window.
28. Drawing different shapes.
29. Using bitmaps.
30. Using fonts.
31. Using metafiles.
32. Using the color palette.
33. Making 3-d dialog windows.
34. Creating custom controls.
35. Making a screen saver.
36. Making nonrectangular windows.
37. Using complex shapes as windows.
38. Creating a file object.
39. Saving different objects in a file.
40. Reading and writing parts of the file.
41. Creating a DDE server.
42. Creating a DDE client.
43. Getting updated data from the server.
44. Executing commands on the server.
45. Getting Drag-and-Drop files.
46. Creating an OLE client.
47. Creating an OLE server.
I hope I didn't bore you listing all of the subjects covered
in the book. It is a great reference book for Borland C++/Turbo
C++ ObjectWindows programmers. I predict that this book won't
spend much time on my bookshelf.
Contacting me online:
CompuServe: 72163,2165
America Online: RodneyB172
You can also contact me at the following Internet addresses:
CompuServe: 72163.2165@compuserve.com
America Online: RodneyB172@aol.com
- 42 -
Creating Help Files
By Todd Snoddy
Like most of the readers of Windows Programmer's Journal, I
enjoy Windows programming. Although I don't currently have the
blessing?/curse? of being a "full time programmer", I do consider
myself to be a dedicated and serious Windows programming
hobbyist. I would like to use the opportunity that WPJ provides
to share with you some of the tools I use to increase my
productivity. As my busy schedule permits, I am hoping to be
able to review different programming utilities in upcoming
issues.
I would like to be able to concentrate mainly on both
shareware and free software. Although I don't have anything
against commercial software, many of the software packages sold
commercially are rather expensive, putting it out of the budget
of most hobbyists or small business users. In addition,
shareware has many advantages besides usually being less
expensive. Many of the shareware products available fill a void
that isn't quite covered by the major software publishers because
they don't take the time to develop something that won't have
mass market appeal. I'll try to point out some of these products
in the future. Also, the quality of modern shareware can often
approach or surpass that of commercial software.
My review topic this month will be Windows help files. I
will cover a few tools that I use to develop my help files with,
and actually present a sample help file that can be used as an
example for your own help files. The tools that I will discuss
are free, and are available from various sources, including
CompuServe and numerous Internet ftp sites.
Before I get started with that however, I would like to take
a brief moment to tell you a little about myself so that you can
get an idea of my perspective. I am currently in the U.S. Army
and stationed at Ft. Riley, KS. I work as a computer
technician/soldier during the day, and in my free time I do a lot
of Windows programming on personal projects. I am almost
entirely self taught, and am competent in C, Pascal and assembly.
I have been programming Windows for around two years now, and C
for about seven years. Prior to that, I programmed BASIC for
several years. As mentioned before, programming is a major hobby
of mine, but I also like to read and play pool, usually 9 ball.
Now to the good stuff. Currently, there is only one vendor
who makes a compiler capable of creating Windows help files, and
that is of course Microsoft. This help compiler takes a file
that contains a special formatting language called Rich Text
Format, or RTF, and converts it into a HLP file that can be used
- 43 -
by the WinHelp engine included with Windows.
Contrary to what Microsoft would like to have us believe,
it's not necessary to use an RTF word processor like their own
(whose name we all know) to create Windows help files. Although
the actual format of the HLP file is considered by many to be one
of the best kept secrets in the industry, there are several low
cost or free utilities available to generate the RTF file
necessary for input to the Windows help compiler.
When I first wanted to create my own help files, I looked at
several utilities to see what was available, especially since I
didn't have several hundred dollars extra to buy an expensive
word processor. The one that I found that was overall easiest to
use is RTFGEN, by David Baldwin. This is an excellent utility
whose use is free, not even shareware. With David's permission,
an older command line version is included with this issue of WPJ,
along with the Pascal source code, although a newer Windows
version is available for free as well. For a small fee, you can
even obtain the source for the Windows version.
RTFGEN's sole purpose is to convert standard ASCII text
files into an acceptable RTF file recognizable by the help
compiler. Using RTFGEN for creating your help files is almost
like using a programming language like C. Many people say that C
is a high level language with the flexibility of a low level
language like assembler. RTFGEN will let you mix straight RTF
code with your help source file for maximum control over the
desired output, or you can use the default settings and just
specify the actual topic information and let RTFGEN handle the
rest. RTFGEN has some special keywords that will actually be
converted into RTF automatically, so you don't have to get your
hands dirty if you really don't want to.
Creating a help file with RTFGEN involves a few steps.
First, you need to plan the layout of the help file. The
importance of this step can't be overstated. A well thought out
layout makes a big difference in the apparent quality of your
product, especially when the user is totally lost and in a blind
panic tries to bring up the online help instead of opening the
manual.
The next step would be to create the help source files. For
this task, I use a free Windows editor called the Programmer's
File Editor by Alan Phillips, but it can be done using any editor
that reads and writes standard ASCII files, like DOS's EDIT
command, or even the Windows Notepad.
The help source files contain the actual help topic
information, including the topic text, keywords, and browse
- 44 -
sequences. These files will have the RTFGEN commands in them,
and may also have RTF statements for added flexibility or
control of the help file. Each help topic will be in a specific
format, and this will be covered shortly.
After creating the help source files, we are almost ready to
compile them into a HLP file. However, before this can be done,
the source files need to be processed by RTFGEN. This procedure
converts the special RTFGEN commands into RTF automatically. If
you have any actual RTF statements in the source file along with
the rest of the text, it will be left alone and passed on
unmodified. This allows the ultimate in customization of the
help file.
There is one final step required before actually compiling
the help file. This is to create a Help Project file, or HPJ
file. This file specifies certain options to the help compiler,
including the source files for the help file, bitmaps, and any
help macros used. I won't cover this subject in more depth here
as it is documented in other places like the help compiler
documentation. By the way, another excellent source of
information on the format of the HPJ file as well as help file
creation in general is the Help Author's Guide. This is a help
file created by someone at Microsoft but is not supported
officially by them. It is usually distributed as HAG.ZIP and can
be found on Compuserve and Internet ftp sites such as
FTP.CICA.INDIANA.EDU. It contains a complete reference for the
RTF command syntax.
That's all you need to create a help file. That's all?
Well, as mentioned before, the help topics must be in a specific
format. David Baldwin's documentation for RTFGEN does a good job
of explaining how to use it, but I'll briefly cover some of the
highlights.
First, every RTFGEN command is started by a \ character, and
most are terminated by a ` character. For example, to specify a
topic title, the command would be:
\title Selecting Background and Foreground Colors.`
This statement specifies what text will be displayed in the
Go To list box when the user tries to search for the keywords
"background colors, selecting". See Figure 1 for an example (see
the Help file - Ed.).
In Figure 1, the text shown in the upper list box is
specified by the topic keywords. To get the currently selected
text to be used for the shown topic, the following statement
would be used:
- 45 -
\keyword background colors, selecting`
Every topic in the help file has to have some way to
distinguish it as an individual topic and separate it from every
other topic. This is known as the topic name. The topic name is
never actually seen by the person viewing the help file, but is
instead used as an internal marker to identify the individual
topic to the help system. In RTFGEN, the topic is specified by a
statement like:
\topic SelectingColors`
In the help source file, all of the above statements would
be grouped together in the topic heading. The topic heading is
separated from the rest of the topic text by using a minimum of
five = characters at the end of the heading. In practice, this
would look like:
\topic SelectingColors`
\title Selecting Background and Foreground Colors.`
\keyword background colors, selecting`
=====
This is the sample topic text for selecting a background and
foreground color.
------
In the above example, the actual topic text follows the =
character, and is terminated by five or more - characters. This
is how RTFGEN determines the separation between topics, and based
on their position, it will generate the appropriate RTF code for
the help compiler.
RTFGEN can be used to generate hypertext jumps or popup text
as well. For example, if you have some concepts in a topic that
the user may not be familiar with, you can designate certain
keywords in the topic that the user can select on that brings up
a popup window with a more detailed description or definition of
the keyword.
To define a hypertext jump with RTFGEN, the text that will
be selected by the user to jump to the other topic will be
enclosed in brackets along with the name of the topic to be
jumped to. For example, Figure 2 shows a typical display of a
hypertext jump (see the Help file - Ed.).
To duplicate this affect using RTFGEN, the following
statement would be used in your help source file.
- 46 -
[Choosing Fonts and Font Sizes:ChoosingFonts]
In this example, the ChoosingFonts will represents the topic
name that will be jumped to when the hypertext link is selected.
If instead you wanted to use a hypertext popup window for a
definition, then you would use two sets of brackets. An example
of this is shown below (see the Help file for Figure 3 - Ed.).
picture elements ([[pels:pels_def]]).
Since I am a firm believer in the concept of explaining
something and then giving an example of how to do it, a sample
help source file is included, along with all of the required
files to build it, except for the actual help compiler. You will
need the Windows 3.1 version of the help compiler to build the
sample. After studying the sample files, as well as the
documentation for RTFGEN, you should be well on your way to
creating your own help files.
To compile the sample into a help file, use the command:
SAMPLGEN sample
This will invoke the SAMPLGEN.BAT batch file to convert your
source file into RTF, and if everything goes ok, it will then run
the help compiler.
There are many ways to customize your help files, and I may
cover some techniques in future issues of WPJ as time permits,
but if you want to learn some techniques on your own, then beg,
borrow, steal, or otherwise obtain a copy of the previously
mentioned Help Author's Guide. You can't beat the price since
it's free, and it is full of useful information on RTF and
creating help files.
I welcome any comments anyone may have on this article, as
well as suggestions for other products to review in the future.
I can be reached on America Online at TSnoddy, on Compuserve at
71044,1653, and on the Internet at tsnoddy@nyx.cs.du.edu.
Although I'm very often quite busy, I will try to respond to any
messages.
Also, if you do decide to start using RTFGEN, please contact
the author and let him know what you think of it. The world
needs more software authors who are willing to make useful tools
like this available for free, and when someone makes a
contribution like that, receiving feedback from users provides a
reason to produce other useful utilities too. The author, David
Baldwin, can be reached on Compuserve at 76327, 53.
- 47 -
Gdi See Gdi Do
Graphics Animation in Windows
By Bernard Andrys
In this series of articles, I will discuss the writing of a
space invaders style game for Windows 3.1 in C. The articles
will be directed to people who understand programming (such
BASIC) and know a little about Windows but are new to Windows
programming in C.
Experienced C programmers will immediately recognize the
poor style of my code. In some cases, I will be doing this on
purpose. If a reader is new to C AND Windows, they're probably
going to have just as much of a problem following the C code as
trying to follow the Windows specific areas of programming. (I
know that I sure had problems learning both at the same time.)
As a result, I will try to stick with easy to understand syntax
like "x=x+1" instead of the C shorthand "++x". Later articles
will use more of the shorthand notation as I and the readers get
comfortable with C. The other reason for the bad style is that
my coding philosophy is "Well it works... so I guess its good
enough." This doesn't mean, however, that I don't want to learn
how to write C better. If you are an experience C programmer,
then please drop me a line about how I could improve my
programming style.
What will be different is how we go about writting the
program. I think that the best way to learn programming is to
write a working program, no matter how poorly written, and then
refine it as you learn more. So I will start with the standard
"do nothing" windows program that just creates a window. Future
articles will take this basic program and add the missle base,
missle firing, invaders, invaders firing, and of course the ufo.
From this simple program we add features and fix problems in
future articles. Some of the improved features will be 256 color
bitmaps and digitized sound. A couple of the problems that we
will need to deal with are friendly multitasking and the
efficient drawing of bitmaps. Once the game is finished, I would
like to write about converting the program to C++.
If you are new to Windows programming in C then before you
go on to read the program below, you'll need two books as
references. The first is "The C Programming Language" by Dennis
Ritchie and Brian Kernigan. The second is "Programming Windows"
by Charles Petzold version 3.0. These two books have all the
basics of C and Windows programming in C.
If you are knowledgeable about Windows programming in C then
check out the "Duncan" format of my Windows C code. There's no
case switch statement for processing messages. An array of
- 48 -
structures contains the list of messages that our program will
respond to along with the function that is called for each
message.
A final note before the program starts: This program was
written using Quick C for Windows. As far as I know, I didn't
write anything that was compiler dependant (except the make file,
but then I didn't write it, Quick C wrote it for me). If you
have any problems compiling it, drop me a line so I can remove
any compiler dependancies in future articles. I have been
meaning to purchase Turbo C++ for Windows and/or Visual C++ so
expect make files for one or both of those compilers in the near
future.
// START OF PROGRAM LISTING
/* invid.c */
/*
There are two types of ways of adding comments into C source
code.
The first way to add a comment is to put a / * before the
comment
and a * / at the end of the comment. (I had to add a space
between the / and * because otherwise the C compiler would think
that
it really was a comment. You can't put comments inside of
comments.)
The other way to add a comment in your source code is to put a //
at
the start of the comment. The comment will start at the // and
go to
the end of the line.
*/
#include <windows.h>
#include "invid.h"
/*
The #include statement allows you to include another file in your
source code. This saves the hassle of cutting and pasting the
code in. The file that you include is put inside a pair of < >
signs or a pair of " ". Using < and > tells the compiler to
search through the path defined by an INCLUDE environment
variable (use SET INCLUDE = path to include files). If you use "
" instead, it tells the compiler to only check the current
subdirectory. You normally use < and > for files that come with
with the compiler. Quotes are normally used for your own include
files. Every Windows program needs to include <windows.h>. C by
- 49 -
itself doesn't have any functions for handling character
input/output, character strings and other basic functions.
Therefore most C programs will also include <stdio.h> and
<stdlib.h> to have a bunch of functions to work with.
My own file "invid.h" contains my global variable declarations
and function prototypes. You should already know that a global
variable is a variable that all functions can share. In C, a
global variable is created by declaring a variable above the
function main() ( or for Windows, the function WinMain() ).
Normally you put your global variables at the top of your program
after the #include files (or in your own #include file). Global
variables should normally be avoided because they make your
program larger and make your code harder to manage as it grows.
Function prototypes are a list of the functions that are in your
program. A function prototype includes the data types that the
function will return and the data types of the parameters of the
function. If you were to write a function called DrawPixel(),
you would write:
int DrawPixel( int x, int y)
{
...the function's code goes here
}
Int is the data type that the function returns. DrawPixel is the
name of the function. X and y are the function's arguments with
int x and int y saying that the arguments are of data type
integer. To make a function prototype just copy:
int DrawPixel (int x, int y)
from your program, paste it at the top of your program (or put it
in a separate #include file like I did with invid.h), and add a
semi-colon to the end of the line.
The function prototype for function DrawPixel(x, y) is:
int DrawPixel (int x, int y);
The function prototype tells the compiler about your DrawPixel
function before its used. This lets the compiler check to see
that you use your DrawPixel function with the right data types.
*/
struct decodeWord {
WORD Code;
LONG (*Fxn)(HWND, WORD, WORD, LONG); };
struct decodeWord messages[] = {
- 50 -
WM_CREATE, DoCreate,
WM_DESTROY, DoDestroy, } ;
/*
I have used a different method for having a Windows program
respond to messages than what is normally seen in Windows
programming books. It is a method that I first saw used by Ray
Duncan in PC Magazine. Instead of a large switch case statement,
it uses an array of structures that list the messages that your
program will respond to and the function that is called when that
message is received. (I'm going to skip covering the details of
arrays and structures in C for now. Otherwise, I wouldn't have
anything to write about in future articles!) This format allows
you to easily add new messages without having to deal with a long
and messy switch case statement. To use the Duncan
format, simply add the message that your program will respond to,
a comma, the name of the function that will be called when you
receive that message, and then another comma. In the case above,
if we wanted to have our program respond to a WM_SYSCOMMAND
message (this is the message sent when you click on the system
menu of a window) we would make the following change to the lines
above.
struct decodeWord messages[] = {
WM_CREATE, DoCreate,
WM_SYSCOMMAND, DoSysCommand, // the Windows Message, Our
Function
WM_DESTROY, DoDestroy, } ;
Now when Windows sends a WM_SYSCOMMAND message to our program,
our program will run the function DoSysCommand. (I could have
called the function anything I wanted.) We would then go to the
bottom of the program code and write the function DoSysCommand().
*/
int PASCAL WinMain (HANDLE hInstance,
HANDLE hPrevInstance, LPSTR lpszCmdParam, int nCmdShow)
/*
This is start of the Windows program. (A regular C program
starts with a function called main().) The keyword PASCAL tells
the C compiler that this function will be called with the PASCAL
calling convention. This means that the code that is written
inside of this function has access to the actual variables that
are passed as parameters (i.e., hInstance, hPrevInstance,
lpszCmdParam, nCmdShow). A normal function in C doesn't let you
change the original variable, it gives the function a copy of the
variable to play with.
If the keyword PASCAL was NOT used above, then hInstance would
- 51 -
contain a duplicate of the number that Windows uses to keep track
of which copy of this program we have in memory. (In Windows, we
can have the same program loaded several times.) If we changed
the value of hInstance, it wouldn't change what Windows thinks
hInstance contains. It would only change what our function
thinks hInstance is.
Using the keyword PASCAL means that if we change the value of
hInstance, it will change what Windows thinks the value of
hInstance is as well. This can be very dangerous so you should
use care when dealing with the parameters to functions that are
declared using the PASCAL keyword. So why use the PASCAL
keyword? There are three reasons for using the PASCAL keyword.
The first is that when writing parts of a program, it's often
easier to use the keyword PASCAL to get access to the actual
arguments than to have to deal with creating a group of global
variables or use a pointer to the arguements to get to the
original variable. The next reason is that the PASCAL calling
convention for functions makes programs a little smaller and
makes them run a little faster. The last reason is that any
function that Windows will call in your program must be declare
as a PASCAL function. A simple Windows program only has the
functions Winmain() and WinProc() called by Windows. Winmain()
is the start of your program and WinProc() is the name usually
given to the function that Windows calls when it has a message
for your program. A function that Windows calls inside of your
program is called a callback function and all callback functions
must use the PASCAL keyword. I won't spend any time here
discussing callback functions because using them is a lot more
messy than just adding the keyword PASCAL to the function's
declaration.
*/
{
MSG msg ;
/*
Declare a variable msg of data type MSG (message). This will be
the variable that will hold any messages that are sent to our
program.
*/
if (!hPrevInstance) InitApplication(hInstance);
/*
hPrevInstance is a number that Windows passes to WinMain as a
parameter when run. The number identifies the previous copy of
our program that is currently running.
If this is the only copy of the program running, hPrevInstance
contains 0. In C, the if() statement executes the next statement
if what is in the parenthese's is true (and in C, true is
anything except 0). The ! operator in C performs a logical NOT
- 52 -
operation. A logical NOT operator changes a 0 to a 1 or a 1 to a
0. Therefore if hPrevInstance is 0 (no previous instance
running) then !hPrevInstance equals 1 (True). So if
!hPrevInstance is true then "if (!hPrevstance)" runs the next
statement which is InitApplication(). Once you've seen this a
couple of times, you'll automatically translate a statement like:
if (!hPrevInstance) InitApplication(hInstance);
to mean:
If there is not a previous instance of our program in memory then
run the function InitApplication().
InitApplication is a function that we will write to setup the
information for this window that can be shared between all copies
of our program that are running.
*/
InitInstance(hInstance, nCmdShow);
/*
Call InitInstance(). This is the function that sets up the
information for this program's window that is unique for this
particular copy in memory.
*/
while (GetMessage (&msg, NULL, 0, 0))
{
TranslateMessage (&msg) ;
DispatchMessage (&msg) ;
}
/*
This is the message loop that keeps our program running. Your
program runs in this infinite loop as long as the function
GetMessage() returns TRUE (In C, TRUE is anything other than 0.)
The GetMessage() function returns FALSE if a WM_QUIT message is
received. This means that the while loop will stop if our
program receives a WM_QUIT message and our WinMain() program will
continue to the lines below.
*/
return (msg.wParam) ;
}
/*
This is the end of our Program. What follows are the functions
that WinMain calls and the functions that we call when we receive
a message as defined at the top of this program in the array
messages[].
*/
BOOL InitApplication(HANDLE hInstance)
/*
InitApplication will be the data that is shared between all
- 53 -
instances of our program.
*/
{
WNDCLASS wndclass;
wndclass.style = 0 ;
wndclass.lpfnWndProc = WndProc ;
/*
For this program WndProc() will be the name of the function that
windows calls when it has a message for our program.
*/
wndclass.cbClsExtra = 0 ;
wndclass.cbWndExtra = 0 ;
wndclass.hInstance = hInstance ;
wndclass.hIcon = LoadIcon (hInstance, szAppName) ;
/*
Load our icon for this program. szAppName is a variable defined
in invid.h. (this is just a short cut so that if I use this code
to make a different program, I can just change the variable
szAppName in invid.h instead of having to go through all my
source code to see where I used the name of my program.) Take a
look at the file invid.rc for the rest of the information about
how you add an icon.
*/
wndclass.hCursor = LoadCursor (hInstance, IDC_ARROW) ;
wndclass.hbrBackground = GetStockObject (BLACK_BRUSH) ;
/*
the background of our window will be painted black.
*/
wndclass.lpszMenuName = NULL ;
wndclass.lpszClassName = szAppName ;
return(RegisterClass (&wndclass));
}
BOOL InitInstance(HANDLE hInstance, WORD nCmdShow)
/*
The InitInstance() function will initialize the data that will be
unique for a particular copy of the program that is in running.
*/
{
hWnd = CreateWindow (szAppName,
"Invid",
DS_SYSMODAL | WS_CAPTION | WS_SYSMENU | WS_MINIMIZEBOX,
/*
The flags DS_SYSMODAL etc. are constants defined in windows.h
that tell windows how the window should look. An | symbol is
used between the flags because it is the C operator for logical
- 54 -
OR. If you have the binary numbers 1000 and 0100 then 1000 |
0100 = 1100. So by putting an | symbol between each flag, we
combine the different flags into a single number that Windows
uses to figure out how the window should look.
*/
CW_USEDEFAULT, CW_USEDEFAULT,
WINDOW_WIDTH, WINDOW_HEIGHT ,
/*
A normal Windows program would use the Windows constants
CW_USEDEFAULT above. However, since this is a graphic game whose
window cannot be resized (see DS_SYSMODAL flag above), we are
going to define precisely the size of the window in pixels.
WINDOW_WIDTH and WINDOW_HEIGHT are constants that I've defined in
"invid.h".
*/
NULL,
NULL,
hInstance,
NULL) ;
ShowWindow (hWnd, nCmdShow) ;
UpdateWindow (hWnd) ;
return (TRUE);
/*
In <windows.h> TRUE is defined as 1 and FALSE is defined as 0.
*/
}
long FAR PASCAL WndProc (HWND hWnd, WORD wMsg, WORD wParam, LONG
lParam)
/*
This is the function that Windows calls whenever it has a message
for our window. WinProc goes through each item in the array
messages[] (which was defined at the top of our program) and
checks to see if the message it received matches the message in
the array. If the message does match, then it runs the function
that goes with that message.
*/
{
int i;
for(i=0; i < dim(messages); i++)
{
if(wMsg == messages[i].Code)
return((*messages[i].Fxn)(hWnd, wMsg, wParam,
lParam));
}
return(DefWindowProc(hWnd, wMsg, wParam, lParam));
}
- 55 -
LONG DoCreate(HWND hWnd, WORD wMsg, WORD wParam, LONG lParam)
/*
This is where we will load our bitmaps and initialize some of our
variables. The program doesn't do anything yet so there is
nothing to create.
*/
{
return 0;
}
LONG DoDestroy(HWND hWnd, WORD wMsg, WORD wParam, LONG lParam)
/*
The program doesn't create anything yet so there is nothing to
destroy before the program exits.
*/
{
PostQuitMessage (0) ;
return 0 ;
}
// END OF PROGRAM LISTING
So that's it for this month. Don't you just hate all these
programs that do ABSOLUTELY NOTHING. I sure do! Unfortunately
this article is already a bit long so you'll have to wait until
next month before we turn this "do nothing" Windows program into
a nifty little space invaders style game. If you'd like a peek at
where I'm going with this article, run peek.exe. (You will
probably need to play with the delay= value in the invid.ini file
that is created in your Windows subdirectory after the program is
run the first time.)
About the author:
(short version)
SWM 25 ISO SWF. Enjoys long walks, sun, surf, skiing.
(long version)
Bernard Andrys is a network engineer for CSI, an ISDN hardware
and software development company. He holds a bachelor's degree
in Mechanical Engineering from the University of Maryland at
College Park. He can be reached through the Internet at
andrys@csisdn.com or on the Windows Programmer's Journal BBS.
- 56 -
Book Review: Windows++
By Philip Sloss
I started trying to write Windows programs the old-fashioned
way -- copying examples from magazines and books. This is still
my preferred method when it comes to things I don't understand
clearly. ("Things I Don't Understand Clearly" -- this is such an
all-encompassing topic, that brevity and disk space prevent me
from fully explaining it; however, you will find examples
sprinkled throughout the rest of this article.) I do like
trolling the book stores looking for interesting examples on
Windows programming, and one of my first finds was a book called
"Windows++."
"Windows++: Writing Reusable Windows Code in C++"
Often, you can find out an important thing in the preface to
a book or article: what the book is about. The preface of
"Windows++" states that it is "essentially a recipe for building
a C++ class library interface to Windows." What really sold me
on this book was that the author, Paul DiLascia, actually *shows*
how to do this. Instead of finishing the book with a jumble of
ideas that you have to put together, Paul goes the whole twelve
yards, and shows how to put the library together. The end result
is that "Windows++" is not just a book on how to build a class
library, it is a class library.
Armed with (or, rather, disarmed with) a general naivete
about Windows programming and a lack of enthusiasm for writing
long repetitious code, I decided to use the Windows++ library to
write Windows programs. The book and the library are mutually
beneficial: not only does the library serve the book, but the
book is the perfect reference for the library! When I'm having
problems, I can see how something was implemented in the source
code, and read a detailed explanation of it in the book. The
book is also educational on Windows programming in general.
A couple of points: First, class libraries are very helpful
and can be of great benefit. Microsoft's Visual C++ is
permanently entwined in their class library, the Microsoft
Foundation Class Library (MFC). MFC is the primary reason that
Visual C++ is so useful. Second, and just as important, I think
that one still has to spend time learning Windows the "C" way.
There is no way around it, in my smug, slight opinion.
For example, I'll use someone I'm on speaking terms with --
me. Shortly after purchasing "Windows++," I went and got the
bible -- "Programming Windows" by Charles Petzold. What happened
with me was that I quickly began asking a lot of questions like
"Well, how does this work?" (Compiler errors will cause one to
- 57 -
speak out loud to no one in particular.) It helped to know how
Windows++ calls were mapped to the Windows API, because then I
had some clue as to what to expect. The general consensus is
that Mr. Petzold's book is the best there is on Windows
programming fundamentals -- I agree.
In other words, while I'm led to believe that class
libraries like MFC and Windows++ can greatly simplify one's work,
I still think most beginning Windows programmers are in for a
future showdown with the Windows API.
"Hello, World"
The proliferation of the phrase "Hello, World" has extended
beyond computer programming and now has completely taken over the
information age. The standard introduction to just about anything
-- renting a car, sky diving without a parachute (also known as
sky jumping or sky plunging), underwater basket cooking -- is a
square with the words "Hello, World" in the middle of it. On
your first parachute-less sky dive, this is a good thing to aim
for, as it will allow the authorities to easily locate your
remains.
Windows programming adds a system menu, title bar, and
maximize/minimize buttons to the "Hello, World" square. In
addition, it requires a rather detailed preparation of structures
and calls to functions with 423 parameters. It is also,
apparently, the basis for beginning a Windows class library.
In C, the "Hello, World" program looks like this (adapted,
with few alterations, from Petzold's "Programming Windows"; those
of you who have seen this countless times before, please try not
to become physically ill):
////////////////////////////////////////
#define STRICT
#include <windows.h>
LRESULT CALLBACK _export WndProc (HWND, UINT, WPARAM, LPARAM);
int PASCAL WinMain (HINSTANCE hInstance, HINSTANCE hPrevInstance,
LPCSTR lpszCmdParam, int nCmdShow)
{
HWND hwnd;
MSG msg;
WNDCLASS wndclass;
static char szAppName[] = "Hello";
if (!hPrevInstance) {
wndclass.style = CS_HREDRAW | CS_VREDRAW;
- 58 -
wndclass.lpfnWndProc = WndProc;
wndclass.cbClsExtra = 0;
wndclass.cbWndExtra = 0;
wndclass.hInstance = hInstance;
wndclass.hIcon = LoadIcon (NULL,
IDI_APPLICATION);
wndclass.hCursor = LoadCursor (NULL, IDC_ARROW);
wndclass.hbrBackground =
(HBRUSH)GetStockObject(WHITE_BRUSH);
wndclass.lpszMenuName = NULL;
wndclass.lpszClassName = szAppName;
RegisterClass (&wndclass);
}
hwnd = CreateWindowEx(NULL, // extended window style
szAppName, // window class name
"The Hello Program", // window caption
WS_OVERLAPPEDWINDOW, // window style
CW_USEDEFAULT, // initial x position
CW_USEDEFAULT, // initial y position
CW_USEDEFAULT, // initial x size
CW_USEDEFAULT, // initial y size
NULL, // parent window handle
NULL, // window menu handle
hInstance, // program instance handle
NULL); // creation parameters
ShowWindow (hwnd, nCmdShow);
UpdateWindow (hwnd);
while (GetMessage (&msg, NULL, 0, 0)) {
TranslateMessage(&msg);
DispatchMessage(&msg);
}
return msg.wParam;
}
LRESULT CALLBACK _export WndProc (HWND hwnd, UINT message,
WPARAM wParam, LPARAM lParam)
{
HDC hdc;
PAINTSTRUCT ps;
RECT rect;
switch (message) {
case WM_PAINT:
hdc = BeginPaint (hwnd, &ps);
GetClientRect (hwnd, &rect);
DrawText (hdc, "Hello, Windows!", -1, &rect,
- 59 -
DT_SINGLELINE | DT_CENTER | DT_VCENTER);
EndPaint (hwnd, &ps);
return 0;
case WM_DESTROY:
PostQuitMessage (0);
return 0;
}
return DefWindowProc (hwnd, message, wParam, lParam);
}
Again, all this gets you is a window that can be closed,
with a line of text in it (although it does help pad out this
article). And, for the most part, all programs have to have this
code somewhere in them. This is where the class library comes
in. A class library does repetitive things like filling the
WNDCLASS structure and calling CreateWindowEx() with the correct
parameters for you. You only have to define the styles and
parameters specific to your application.
What "Hello, World" looks like using "Windows++"
For purposes of introduction, one of the first sections in
the book shows what the Windows++ version of "Hello, World" looks
like. It reduces the long, C-style program to this:
#include <wpp.h> // main Windows++ header file
APPCLASS HelloWin : public WPMainWin {
public:
HelloWin() { createWin("Hello"); }
void paint(WPPaintStruct &ps) {
WPRect clientArea = this;
ps.drawText(clientArea, "Hello, Windows++.",
DT_SINGLELINE | DT_CENTER | DT_VCENTER);
}
};
void WPApp::main()
{
mainWin = new HelloWin;
run();
}
As the "Hello, World" example is a do-nothing program, this
program has very little code -- a nice correspondence. For a
detailed explanation of how this is orchestrated, I strongly and
emphatically suggest purchasing "Windows++."
- 60 -
What's the Difference?
What separates "Windows++" from some of the other books on
C++ Windows programming is that "Windows++" is primarily about
how the library fits together. Often, all one gets is the source
code to a class library at the back of the book and/or on disk
and instructions on how to use it. "Windows++" shows how anyone
can build their own library -- it discusses some of the issues
one has to deal with, and some of the standard and extended
features one should consider incorporating in a class library.
At the same time, there are several instructive examples of real
Windows programs written with the Windows++ class library.
And finally, in the last chapter, the reader is shown how to
put the class library into a dynamic link library (DLL), and the
issues and problems that any library has to deal with to put C++
classes into a DLL.
"How About a Sudden Ending?"
So, summing up: "long, redundant code -- bad, reusable C++
Windows classes -- good." And a good book on how to make a C++
class library for Windows is "Windows++."
About the Author: An Unauthorized Autobiography of Philip Sloss
Born from a crude genetic experiment mixing the genes of a
human sperm with tree sap, Mr. Sloss has led a far from legal
existence since his grudging inception into the world. After
being developed in an artificial womb, he was raised in a large
culture of athlete's foot fungus. Mr. Sloss was accepted into
the aerospace engineering department at San Diego State
University, and in May of 1992, having advanced to the rank of
graduate, he left college and began a career in unemployment. He
is now under under psychiatric observation for "behavior
unsuitable to inert objects." Hospital officials report that Mr.
Sloss' main hobby is spotting lint.
- 61 -
Getting in touch with us:
Internet and Bitnet:
HJ647C at GWUVM.GWU.EDU -or- HJ647C at GWUVM.BITNET (Pete)
GEnie: P.DAVIS5 (Pete)
CompuServe: 71141,2071 (Mike)
WPJ BBS (703) 503-3021 (Mike and Pete)
You can also send paper mail to:
Windows Programmer's Journal
9436 Mirror Pond Drive
Fairfax, VA 22032
U.S.A.
In future issues we will be posting e-mail addresses of
contributors and columnists who don't mind you knowing their
addresses. We will also contact any writers from previous issues
and see if they want their mail addresses made available for you
to respond to them. For now, send your comments to us and we'll
forward them.
- 62 -
The Last Page
By Mike Wallace
I got this in the mail a few days ago from Mike Strock...
Here is a wonderful quote I think you will just love.
It comes from the Journal American, page d6, in the 2nd
article, 3rd paragraph in the 4th column.
"Electronic bulletin boards are fast becoming a popular
way to get copies of programs for free, but the
practice is illegal. Bulletin boards are computers
that can be reached by other computers via modem. Most
of them require the new users to upload an illegal
program onto their machines (thereby incriminating
themselves) before they are allowed to down-load other
programs for free, said Alison Gilligan, international
anti-piracy specialist at Microsoft."
Bulletin boards have been blamed for spreading illegal
copies of software almost since the day they started appearing.
Software companies have long complained their very existence is
in danger due to these BBSs spreading free copies of their
software. Is this fair to the majority of BBSs? No, I don't
think so. I've been on a lot of boards, and most of the ones
I've seen aren't doing anything illegal. Maybe you need to know
someone to get onto the illegal boards, but that doesn't
legitimize the above quote from Microsoft. The trend in this
country to blame others for your own "bad luck" gets out of hand
sometimes. Say you start up a commercial software company and
nobody buys your app. Does that mean everyone has a free copy
obtained from a BBS? Maybe, but there's also a chance your
program isn't worth buying, or people don't even know about it.
It's hard to buy every commercial program you want when each one
may cost $200-300, or more. The software companies say this is
to recoup their losses from illegal (unpaid) copies floating
around. I think this may have the opposite effect - who can
afford to buy software that's so expensive? Now, I'm not
condoning software piracy. Besides being illegal and immoral, I
think everybody loses when nobody's paying for their software.
The concept of "shareware" should be extended to commercial
software. Will this ever happen? Never mind whether it's even
feasible. I would probably own more software if I could try it
out first and then pay for it if I decide I'll use it. But as it
is, I can't afford to pay big bucks for a program that I may
never use once I've tried it out. So, the result is there's
probably a lot of good software for sale that would make my life
easier, but I'll never know about it. The same can probably be
said for a lot of computer users. Hmmmm...sounds like
- 63 -
everybody's losing again.
A lot of these companies only appear to be in the business
of manufacturing software so they can make a lot of money, and I
think the people putting out quality software are. Take Borland,
for example. They put out a lot of good programs, and they're
reaping the benefits. Microsoft too. Why they're complaining
about BBSs is beyond me. Let's look at one BBS and see how
Microsoft has benefited from it: the WPJ BBS. Besides carrying
WPJ, there's a lot of other software (all legal) available from a
CD-ROM that can be accessed by anybody that calls our BBS. Most
of it is devoted to products produced by Microsoft (e.g.,
Windows), and we're just part of a much larger network of BBSs
all over the world supporting Microsoft. True, a lot of
commercial software companies live off Microsoft by writing apps
helping people run Windows, but it goes both ways: Microsoft
benefits by having so many companies support their products with
add-on programs or whatever. If I was going to choose an
operating system for my computer, one thing I would look at would
be the third-party support, and Microsoft is the overwhelming
leader in that area for IBM-compatible PCs. So, to hear
Microsoft complaining about BBSs is more than a little annoying.
Sounds like it's biting the hand that feeds it.
- 64 -