OS/2 Professional Developers Kit 1992 November

home *** CD-ROM | disk | FTP | other *** search

/ OS/2 Professional Developers Kit 1992 November / Disc01 / Disc01.mdf / os2bbs / os2prg.cfo < prev next >

Wrap

Text File | 1992-10-31 | 94.6 KB | 2,379 lines

This forum is supplied for demonstration purposes only and is accessible via the IBM OS/2 BBS system. What you see below is the forum as it existed on 10/30/92. The IBM OS/2 Bulletin Board System (BBS) is IBM's new electronic support offering for OS/2 Version 2.0. With OS/2 BBS you can electronically review OS/2 technical information, submit program defects to IBM, and receive information on how to solve problems. Users of the OS/2 BBS not only include OS/2 customers, but the OS/2 expertise of the entire internal IBM community worldwide. OS2BBS makes all this available via a local or toll free number in most major cities. To request a complete OS/2 BBS Information Package detailing the OS/2 BBS offering and including pricing and registration/access information, please call: 1-800-547-1283. ===== OS2PRG CFORUM created at 01:50:23 on 92/04/08 GMT (by HOPKINS at AUSVM8) IBM UNCLASSIFIED ---------------- The information in this forum is considered by IBM to be external and may be shared by everyone who is given access. This forum is shared with the internal IBM employee conferences. Forum: OS2PRG CFORUM Forum subject: OS/2 Base Programming and System Routines This FORUM is provided as a conferencing tool to promote discussion of issues, questions, and problems relating to OS/2 Base Programming and System Routines. Contributions must not contain confidential or proprietary information. Each contributor agrees, in placing material in this forum, that IBM and other users are free to prepare derivative works based upon it, and to use, reproduce, perform, display and distribute inside and outside their company such material and such derivative works, without obligation or charge. IBM is not responsible or liable for the use, copying, or distribution of materials placed on this forum, and has therefore no obligation to control such acts. All subscribers are required to abide at all times by the conference rules described in the Bulletin Board Rules under "News and Announcements". All IBM personnel are required to abide at all times by the IBMPC RULES and the IBM Business Conduct Guidelines. IBM personnel will read the CUSTOMER FORUM RULES in IBMPC RULES before contributing. Contributions violating these rules will be deleted by the CFORUM Owner or their delegate, and in some cases by an IBM conference Administrator. The OS2BBS CFORUM is where anyone may raise concerns regarding rules violations found in this and other CFORUMS. The IBM Owner of this forum is: Name: Kenneth M. Hopkins, OS/2 Support Services Mailing Address: IBM Manufacturing and Development 11400 Burnet Road Internal Zip 2900-042/1D003 Austin, Texas 78758 IBM Node/Userid: AUSVM8/HOPKINS ----- OS2PRG CFORUM appended at 22:02:43 on 92/05/05 GMT (by IL80988 at OS2CUST) Subject: comm port address does anyone recall the comm port address for programming ? ----- OS2PRG CFORUM appended at 14:07:34 on 92/05/06 GMT (by SHOWPM1 at TPAVMPS2) Subject: comm port address Ref: Append at 22:02:43 on 92/05/05 GMT (by IL80988 at OS2CUST) If you are writing a device driver then the COM address depends on which COM port you want to use. I think that the address for COM1 is 3F8H. I do not know about other COM ports. If you are writing a OS/2 application all you need to use is DosOpen, to open the COM port, DosRead, to read information from the COM port, DosWrite, to write information to the COM port, and use DosDevIOCtl to control the COM port. Joseph Mogavero ----- OS2PRG CFORUM appended at 19:18:59 on 92/05/07 GMT (by SSIA013 at OS2CUST) Subject: Com port address Shouldn't you get the COM port address from 0:400 - 407? If I install a com port and have it address some place else (other than the standard post), then it should be reflected there. pgk ----- OS2PRG CFORUM appended at 14:34:07 on 92/05/11 GMT (by SHOWPM1 at TPAVMPS2) Subject: Com port address Ref: Append at 19:18:59 on 92/05/07 GMT (by SSIA013 at OS2CUST) Yes, That is correct you would look at the absolute address 400 - 406 to find out the port base address. Thank you, I should have added that to my first reply. If you need more information about this I would get the 'Personal System/2 and Personal Computer BIOS Interface Technical Reference', you can order this book from you branch office. The IBM number on this book is S04G-3283-00. You can also get the book 'Micro Channel Architecture Revolution in Personal Computing', ISBN number 0-442-00433-8 this will help a little. I also know that there are some other books out there that talk about the BIOS. Joseph ----- OS2PRG CFORUM appended at 18:00:44 on 92/05/13 GMT (by TEMP19GD at CLTVM1) Subject: Peeking a memory location How does one go about looking at the contents of memory at some memory location under OS/2 2.0? Thanks, Justin Smith ----- OS2PRG CFORUM appended at 16:36:34 on 92/05/15 GMT (by FIECHTNE at BCRVMPC1) Subject: Peeking a memory location Ref: Append at 18:00:44 on 92/05/13 GMT (by TEMP19GD at CLTVM1) You must be able to build a Selector which includes the memory location where you wish to look. Only a device driver can build selectors. Mark Fiechtner ----- OS2PRG CFORUM appended at 02:57:13 on 92/05/19 GMT (by WZ00541 at OS2CUST) Subject: Converting Windows programs to OS/2 via "MIRRORS" We are currently using the MIRRORS libraries to convert our Windows applications to OS/2. We have run into a couple of problems with MIRRORS. I am trying to find the correct forum to ask questions about MIRRORS conversion and report problem. ----- OS2PRG CFORUM appended at 13:09:37 on 92/05/19 GMT (by LECTION at BETASVM2) Subject: Converting Windows programs to OS/2 via "MIRRORS" Ref: Append at 02:57:13 on 92/05/19 GMT (by WZ00541 at OS2CUST) The most correct forum would be OS2WIN CFORUM. Also, if you have any questions about direct conversion of Windows code to OS/2 code, please ask away in that forum! I've programmed Windows apps for IBM for seven years, and I am now doing OS/2. So I can compare the two programming wise to a great extent. David B. Lection ----- OS2PRG CFORUM appended at 13:14:26 on 92/05/19 GMT (by WEEKS at AUSVM2) - ..... OS2PRG CFORUM modified at 15:26:22 on 92/05/19 GMT (by WEEKS at AUSVM2) . Subject: Converting Windows programs to OS/2 via "MIRRORS" Ref: Append at 02:57:13 on 92/05/19 GMT (by WZ00541 at OS2CUST) | Yeah, OS2WIN is probably better. Craig Weeks OS/2 Application Development Center Austin, Texas ----- OS2PRG CFORUM appended at 18:56:47 on 92/05/19 GMT (by KMCCARTH at ENDVM1) Subject: SYS RC4 I am coding in C using PM and multiple threads and I get an ocasional RC 4 on a dosopen. I coded a dossetmaxfh to 40 but I don't have near that amt of files open. Does anyone know why this may be happening. Kevin E. Mc Carthy ----- OS2PRG CFORUM appended at 21:42:51 on 92/05/19 GMT (by UST1013 at OS2CUST) Subject: OS/2 Monitor aaadkfdjkdj kjasjsdjlasdj lksdjlksdfjlkjsdlsd ----- OS2PRG CFORUM appended at 21:44:05 on 92/05/19 GMT (by UST1013 at OS2CUST) Subject: OS/2 Monitor (no text entered) ----- OS2PRG CFORUM appended at 21:47:29 on 92/05/19 GMT (by UST1013 at OS2CUST) Subject: OS/2 Monitor Looking for an OS/2 monitor. Facilities should include a display summary of the active sessions/processes/threads. Would prefer one written in C that is packaged with C source code. P.W. ----- OS2PRG CFORUM appended at 15:30:19 on 92/05/20 GMT (by WWRAGG at SFOVMIC1) Subject: Converting Windows programs to OS/2 via "MIRRORS" Ref: Append at 13:14:26 on 92/05/19 GMT (by WEEKS at AUSVM2) Thanks for taking care of my customer ----- OS2PRG CFORUM appended at 14:59:42 on 92/05/21 GMT (by IL21834 at OS2CUST) Subject: Keyboard Buffer - Want to increase size I have an OS/2 PM application with a Dialog Box containing an entry field; input to this entry field is either keyboard entry or a bar code scanner; the scanner input is coming through the keyboard port. The problem we have is sometimes the scanner sends data to the keyboard buffer faster than the application can process it; the keyboard buffer is full; the PS/2 beeps; and characters are lost. Two questions: 1) What is the current size of the Keyboard buffer? 2) Is there a way for us to increase the size of the keyboard buffer? Thank You Ken Prince - EI DuPont ----- OS2PRG CFORUM appended at 14:24:38 on 92/05/28 GMT (by AOI1075 at OS2CUST) Subject: SYS RC4 Ref: Append at 18:56:47 on 92/05/19 GMT (by KMCCARTH at ENDVM1) Are you opening files across the LAN? If so, I believe you can get a return code 4 from DosOpen because of the configuration settings on the server. J.Farmer ----- OS2PRG CFORUM appended at 18:04:28 on 92/06/04 GMT (by WZ00541 at OS2CUST) Subject: Converting Windows programs to OS/2 via "MIRRORS" Ref: Append at 15:30:19 on 92/05/20 GMT (by WWRAGG at SFOVMIC1) (no text entered) Joel Rosenberger, Attachmate Corporation ----- OS2PRG CFORUM appended at 20:10:17 on 92/06/04 GMT (by UDSS013 at OS2CUST) Subject: Program Abends During Exit Processing I am in the midst of developing a PM application. One of the bugs that I am currently trying to remedy is an abend during the execution the "return" statement in main(). I am getting a SYS3175. I got into the debugger and set a breakpoint on the "return" statement. If I step into the code, I am dropped into the assembler and I abend there during what I presume is exitlist handling. I have coded no exitlist processing of my own. This is a multithreaded PM application and my current theory is that it has something to do with the cleanup of the threads. I do not terminate the threads in my code. I just exit the main routine. I've never had any problems with this under OS/2 v1.x. This is my first 32-bit PM application. I am using C Set/2. Any suggestions? The threads were started using _beginthread. I did not allocate any thread stack, as I understand that OS/2 v2.0 handles that now automatically. This one's got me stumped. Gary Murphy ----- OS2PRG CFORUM appended at 20:50:07 on 92/06/04 GMT (by OCANUC at BETASVM2) Subject: Program Abends During Exit Processing Ref: Append at 20:10:17 on 92/06/04 GMT (by UDSS013 at OS2CUST) From an OS/2 command prompt, try "help sys3175" (if you haven't already). The error text suggests a problem with pointers rather than threads. Secondary threads should be shutdown automatically when the main thread is shutdown. Brandon Booth Advanced Programming - ICSRS ----- OS2PRG CFORUM appended at 12:08:13 on 92/06/05 GMT (by WZ00576 at OS2CUST) Subject: How to turn an APIRET into a string? Pollling through the documentation, I can't find any way to get a character string version of one of the ERROR_ APIRET values. I could type in my own table of strings for them, but it would only be in english. This is for some substrate code that can't make assumptions about Vio versus PM. Is there something I don't know? Benson Margulies ----- OS2PRG CFORUM appended at 13:58:43 on 92/06/05 GMT (by OCANUC at BETASVM2) Subject: How to turn an APIRET into a string? Ref: Append at 12:08:13 on 92/06/05 GMT (by WZ00576 at OS2CUST) I know exactly what you're talking about because I looked for the same thing a couple of years back. I even started on the project, but quit (and for much the same reason you mention). So, if you write the dll, can I have a copy? (not serious;-) Brandon Booth Advanced Programming - ICSRS ----- OS2PRG CFORUM appended at 17:16:45 on 92/06/05 GMT (by IL11901 at OS2CUST) Subject: How to turn an APIRET into a string? Ref: Append at 13:58:43 on 92/06/05 GMT (by OCANUC at BETASVM2) Funny you should mention this at this point in time. We are finishing up a high level runtime API debugger and production code monitor. It reports the error which occurred AND a description of what the error means. This idea was born out of the problem you describe (having to look up every error code) and the concept of centralizing error handling. Message back if you are interested in more information. Steve Dacek - Soft & GUI Inc ----- OS2PRG CFORUM appended at 17:25:44 on 92/06/05 GMT (by IL11901 at OS2CUST) Subject: Program Abends During Exit Processing Ref: Append at 20:10:17 on 92/06/04 GMT (by UDSS013 at OS2CUST) Most likely, you are not allocating enough stack or heap space (My money would be on heap space). Add a lot more in your definition file and see if the problem goes away. The minimum for both should be about 8K each. This has usually solved the problem you describe whenever it has happened to me. Good Luck. Steve Dacek - Soft & GUI Inc ----- OS2PRG CFORUM appended at 18:42:11 on 92/06/05 GMT (by XXNOVA24 at TORVMCOP) Subject: Program Abends During Exit Processing Ref: Append at 20:10:17 on 92/06/04 GMT (by UDSS013 at OS2CUST) Instead of 'return'-ing you can call the DosExit function. I haven't tried it with GA CSET/2 but this should do the job. Geza Szivos NOVACOR Chemicals (403)-290-6583 ----- OS2PRG CFORUM appended at 20:22:48 on 92/06/05 GMT (by OCANUC at BETASVM2) Subject: Program Abends During Exit Processing Ref: Append at 20:50:07 on 92/06/04 GMT (by OCANUC at BETASVM2) One other thing to look at is the size of the thread stack(s). You can (potentially) have three kinds of stacks: the main stack (specified in the def file), secondary thread stacks (specified on the _beginthread call), and a 16-bit stack if you make calls to 16-bit routines (this value is specified in a pragma in IBM C-Set2). Brandon Booth Advanced Programming - ICSRS ----- OS2PRG CFORUM appended at 22:34:35 on 92/06/05 GMT (by PMUELLR at CARVM3) Subject: How to turn an APIRET into a string? Ref: Append at 13:58:43 on 92/06/05 GMT (by OCANUC at BETASVM2) It may be overkill, but you can get text messages for os/2 error codes from the OSO001.MSG and OSO001H.MSG files in \os2\system (2.0). You can even do this in rexx with SysGetMessage()! Patrick Mueller - IBM Cary ----- OS2PRG CFORUM appended at 21:04:13 on 92/06/09 GMT (by ANFUSO at STLVM27) Subject: VIO Problems I have a non-IBM friend who is trying to port an application to OS/2 (1.1+). I'm not a programmer, but I'll pose his problem as best I can: ....................................................... His application will run in a character-mode full-screen OS/2 session. He is concerned about the default colors available. He contends that the system pallette offers 16 colors, but notices that other applications use colors that don't seem to be available when he tries all of the 16 default colors. He is convinced that these other applications may be re-defining the system pallette, but wonders how this is done. He thinks it has something to do with "VIO". ......................................................... I'm not even sure if this is even a good forum to append this question to, since I don't know if PM has anything to do with it. His program is, apparently, NOT a PM program. Any help at all would be appreciated, even if you can steer me in a better direction! Tom Anfuso J63/G331 TL 543 - 4835 ----- OS2PRG CFORUM appended at 21:02:03 on 92/06/10 GMT (by WZ00576 at OS2CUST) Subject: Vio, in 2.0, what's that? The CP documentation makes passing reference to several vio things, including popups and the like. But no where in the $300 box of programming documentation can I find any other information about it. Is There Another Book? Benson Margulies ----- OS2PRG CFORUM appended at 21:06:38 on 92/06/10 GMT (by WZ00576 at OS2CUST) Subject: How to turn an APIRET into a string? Ref: Append at 12:08:13 on 92/06/05 GMT (by WZ00576 at OS2CUST) I saw the reply from someone at Cary referencing some files with strings in them, and a Rexx procedure to poke around therin. I'm finding it hard to believe that OS/2 2.0 lacks some API for fetching messages out of Locale=dependent files. In short, its hard to believe that there isn't a c-callable way to open the system message catalog and fish out the string for a particular APIRET. WHen I signed up for this BBS, the literature said that I could count on responses from the OS/2 support/devo group. Is this the right place to expect that? Benson Margulies ----- OS2PRG CFORUM appended at 11:53:23 on 92/06/11 GMT (by PMUELLR at CARVM3) Subject: How to turn an APIRET into a string? Ref: Append at 21:06:38 on 92/06/10 GMT (by WZ00576 at OS2CUST) I think you're referring to my append. Sorry I only referenced the rexx function SysGetMessage (in the RexxUtil.dll external functions). There is also a function you can use from C - DosGetMessage() (this is the name in 1.3 - it may have changed in 2.0, but I don't have the 2.0 toolkit at home). Note that the original append seemed to want a way to translate a return code, say 8, into the string "ERROR_NOT_ENOUGH_MEMORY". There is a define in the toolkit for #define ERROR_NOT_ENOUGH_MEMORY 8 I don't know how to do this. The SysGetMessage() and DosGetMessage() functions return the text for the message. For instance, here is a session from rexxtry with the 8 message: [E:\] rexxtry REXXTRY.CMD lets you interactively try REXX statements. Each string is executed when you hit Enter. Enter 'call tell' for a description of the features. Go on - try a few... Enter 'exit' to end. say sysgetmessage(8,oso001.msg) SYS0008: There is not enough storage available to process this command. All available storage is in use. ................................................ REXXTRY.CMD on OS/2 say sysgetmessage(8,oso001h.msg) EXPLANATION: If segment swapping is active, the swap file may be full, or an input/output (I/O) error may have occurred on the auxiliary storage device that contains the swap file. ACTION: Do one of the following and then retry the command: 1. Reduce the number of running programs. 2. Reduce the value of the BUFFERS=, TRACEBUF=, DISKCACHE=, THREADS=, RMSIZE=, or DEVICE=VDISK.SYS statement in the CONFIG.SYS file. Then restart the system. 3. Remove unwanted files from the swap file ................................................ REXXTRY.CMD on OS/2 exit As I said before, this may be overkill. Also note that the rexx function is implemented by calling DosGetMessage(). DosGetMessage() can be used to read any suitable OS/2 message file. Patrick Mueller - IBM Cary ----- OS2PRG CFORUM appended at 20:55:58 on 92/06/11 GMT (by JDEEN at AUSVM1) - Subject:Changing File Attributes I am in the midst of programming an application that at one time may need to write/delete files with the read-only or hidden-file file attributes, I would like to DosOpen() and DosSetFileInfo() but this is clearly out because I can not not DosOpen() with write flags on when the file is write protected. Am I missing something or is there a 2.0 api that I can change a read_only to normal file status.. JAMES DEEN ----- OS2PRG CFORUM appended at 00:35:45 on 92/06/12 GMT (by RICHKULP at CARVM3) Subject: Changing File Attributes Ref: Append at 20:55:58 on 92/06/11 GMT (by JDEEN at AUSVM1) Look up DosSetPathInfo. This allows you to change the attributes of a closed file. The only restriction is that nobody else can have the file open (i.e. Read-Only is OK to allow changes). Rich Kulp ----- OS2PRG CFORUM appended at 17:54:58 on 92/06/12 GMT (by DAVISN at HOUVMSCC) Subject: Vio, in 2.0, what's that? Ref: Append at 21:02:03 on 92/06/10 GMT (by WZ00576 at OS2CUST) The VIO and MOU calls (and maybe some others) are available in 2.0 for compatibility with 1.x. They are not available as 32-bit entry points, and are not documented with the 32-bit toolkit. If you need to find out how to use these calls, find a good book on programming OS/1 1.x sorry, Nathan Davis ----- OS2PRG CFORUM appended at 12:44:40 on 92/06/23 GMT (by WZ00576 at OS2CUST) Subject: How to turn an APIRET into a string? Ref: Append at 11:53:23 on 92/06/11 GMT (by PMUELLR at CARVM3) This is just what I wanted to know. Thanks. Benson Margulies ----- OS2PRG CFORUM appended at 12:50:14 on 92/06/23 GMT (by WZ00576 at OS2CUST) Subject: Vio, in 2.0, what's that? Ref: Append at 21:02:03 on 92/06/10 GMT (by WZ00576 at OS2CUST) The response is not informative. When the Control Program reference, particularly on the topic of DosExecPgm and DosStartSession, references Vio popups, is is describing a feature that is no longer meaningful? Is there no 32 bit API to OS/2 session windows at all? Is there a corresponding PM concept the documentation should reference instead? Benson Margulies ----- OS2PRG CFORUM appended at 14:45:03 on 92/06/24 GMT (by LINSUSAN at HKGVM8) Subject: compiling and linking 16-bit OS/2 API programs in 2.0 I have a sample program which calls the AS/400 PC Support API's. This sample program was compiled using MS C 6.0 and OS/2 1.3 toolkit. Therefore these must be 16 bit API's. How do I call these API's in a 32 bit program developed using C Set/2, Toolkit 2.0? Unfortunaly the DLL that came with this API must be a 16 bit DLL. I seem to recall that I can call a 16 bit DLL from a 32 bit program. What changes must I make to the header file? Right now the header contains statements such as : extern int extentry QSNDDTAQ(........... but my compiler doesn't like it. How about any special options I must set to compile and link this? Susan S. Lin ----- OS2PRG CFORUM appended at 14:48:25 on 92/06/24 GMT (by LINSUSAN at HKGVM8) Subject: Minimize a session I have written a OS/2 full screen program to DosStartSession a DOS FS program, after the DOS program finishes running, I do not want to close the session but want to minimize it and put it in the background. What function do I need to make this happen? There is a DosSelectSession to bring a dependent child session to the foreground but what about moving the child session to the background? Susan S. Lin ----- OS2PRG CFORUM appended at 01:51:09 on 92/06/25 GMT (by LINSUSAN at HKGVM8) Subject: compiling and linking 16-bit OS/2 API programs in 2.0 I have a sample program which calls the AS/400 PC Support API's. This sample program was compiled using MS C 6.0 and OS/2 1.3 toolkit. Therefore these must be 16 bit API's. How do I call these API's in a 32 bit program developed using C Set/2, Toolkit 2.0? Unfortunaly the DLL that came with this API must be a 16 bit DLL. I seem to recall that I can call a 16 bit DLL from a 32 bit program. Do I have to do some thunking or what? What changes must I make to the header file? Right now the header contains statements such as : extern int extentry QSNDDTAQ(........... but my compiler doesn't like it. How about any special options I must set to compile and link this? If someone has a short sample to do this, please send it to me. Thanks. - Susan Lin (IBM China/Hong Kong) ----- OS2PRG CFORUM appended at 14:11:54 on 92/06/25 GMT (by OCANUC at BETASVM2) Subject: compiling and linking 16-bit OS/2 API programs in 2.0 Ref: Append at 01:51:09 on 92/06/25 GMT (by LINSUSAN at HKGVM8) The C-Set/2 Users Guide has a chapter on calling 16-bit functions from 32-bit code. The bit of information that helped me the most was changing the function prototype of the 16-bit call from APIENTRY (which equates to _System) to APIENTRY16 (which equates to _Far16 _Pascal). The linker seemed to figure everything else out. You do have to think about the 16-bit stack. This is a pragma in C_SET/2 (pragma stack16). Brandon Booth Advanced Programming - ICSRS ----- OS2PRG CFORUM appended at 16:13:56 on 92/06/25 GMT (by LNK1SM3 at MARTLINK) Subject: compiling and linking 16-bit OS/2 API programs in 2.0 Ref: Append at 01:51:09 on 92/06/25 GMT (by LINSUSAN at HKGVM8) For a 32 bit program to call a 16 bit API in a DLL you must put: APIRET16 APIENTRY16 16BITAPI( This will tell the compiler to put thunking in for you. If this does not work give me some code and I will look at it. You should not to have to change your header at all unless you have you prototype there. By the does any know if 'thunking' means my program is not 'thinking'! Joseph A. Mogavero ----- OS2PRG CFORUM appended at 17:17:40 on 92/06/29 GMT (by WZ00576 at OS2CUST) Subject: Named Pipe Local Host Name? When using named pipes across a network, it would seem fairly important to be able to determine the local host name -- the name that some other machine would put into \\NAME\pathname. Where can I get it? Benson Margulies ----- OS2PRG CFORUM appended at 15:40:37 on 92/06/30 GMT (by XXTCAN06 at TORVMCOP) Subject: OS/2 command from Smalltalk/V PM 1.4. Hi! Does anybody know how to issue OS/2 commands from a Smalltalk/V PM 1.4 program? I'm specifically looking to run 'NET VIEW' commands. Actually, the next step after that will be to refine the methods to use the NET API (which I'm not familiar with at all!). So, as a 2nd question: Does anybody have any experience in making calls to DLLs from Smalltalk? And were can I find the NET API DLL(s) and documentation? Thanks. Pierre Legault Transport Canada 1-613-993-7149 ----- OS2PRG CFORUM appended at 11:47:47 on 92/07/02 GMT (by WZ00576 at OS2CUST) Subject: DosRead of 0, waits? I need to do a read-with-timeout from a perhaps local or perhaps remote named pipe. I don't see a way given the restriction of semaphore binding to local pipes. Benson Margulies ----- OS2PRG CFORUM appended at 03:22:53 on 92/07/03 GMT (by SDETWEIL at DETVMIC4) Subject: Named Pipe Local Host Name? Ref: Append at 17:17:40 on 92/06/29 GMT (by WZ00576 at OS2CUST) See the NetWkStaGetInfo api in the Lan Server manual... It will return the Workstation name, Domain name and User name (if any)..this can't tell you the SERVER name of the pipe tho, as you CAN talk CROSS DOMAIN with Named Pipes PROVIDED you have access rights to the pipe AND are logged on.. Sam ----- OS2PRG CFORUM appended at 03:23:53 on 92/07/03 GMT (by SDETWEIL at DETVMIC4) Subject: DosRead of 0, waits? Ref: Append at 11:47:47 on 92/07/02 GMT (by WZ00576 at OS2CUST) DosPeekNmPipe will tell you any data Exists.. Sam ----- OS2PRG CFORUM appended at 11:39:24 on 92/07/03 GMT (by WZ00576 at OS2CUST) Subject: DosRead of 0, waits? Ref: Append at 03:23:53 on 92/07/03 GMT (by SDETWEIL at DETVMIC4) I need to make a tnread wait for input but not read any imput. The routine you reference won't do that, will it? Benson Margulies ----- OS2PRG CFORUM appended at 11:42:26 on 92/07/03 GMT (by WZ00576 at OS2CUST) Subject: DosRead of 0, waits? Ref: Append at 03:23:53 on 92/07/03 GMT (by SDETWEIL at DETVMIC4) I don't have the lan server API. What I have is the Novell netware requester, which supports Named Pipes, so I'm trying to just use them with the Control Program API. Is this as stupid an idea as I'm beginning to suspect? Does the Lan Server API work with plain Novell Named Pipes? Somewhat on the subject, is there any chance of making semaphores work with remote named pipes. Its mighty inconvienient not to be able to do any sort of wait on them except DosRead. Benson Margulies ----- OS2PRG CFORUM appended at 18:01:59 on 92/07/06 GMT (by A157623 at OS2CUST) Subject: OS/2 and Smalltalk I have interfaced Smalltalk with DLLs and can share that info with you. The NET APIs calls are documented in the SDK for Lan Manager. I have such a book and can give the details on the calls you will be needing. Larry Mason Texas Instruments ----- OS2PRG CFORUM appended at 18:13:08 on 92/07/07 GMT (by IL12431 at OS2CUST) Subject: Changing Keyboard States Is there example code available showing how to change the keyboard (numlock, caps lock, and scroll lock) states? I recall an older IBM IUO utility that changed startup states but I need example code for per session management. Thanks in advance. Dave Scheffer, ARGO Data Resource Corp. ----- OS2PRG CFORUM appended at 21:26:05 on 92/07/07 GMT (by IL12431 at OS2CUST) Subject: Changing VIO font mode Is there example code available showing how to change the operative video mode of an OS/2 VIO window to 132 character mode? Am I able to do so internally or must I externally define the object/window? Thanks in advance. Dave Scheffer, ARGO Data Resource Corp. ----- OS2PRG CFORUM appended at 14:21:19 on 92/07/09 GMT (by DALNK03 at OS2CUST) Subject: Changing Keyboard States Ref: Append at 18:13:08 on 92/07/07 GMT (by IL12431 at OS2CUST) Here's a fragment of code which turns on the NumLock key when the program is activated, and turns it off when the program is deactivated. Hope it helps: MRESULT EXPENTRY WindowProc (HWND hwnd, USHORT msg, MPARAM mp1, MPARAM mp2) { static BOOL Toggled; static BYTE KeyStateTableì256┘; switch (msg) { case WM_ACTIVATE: case WM_SETFOCUS: { SHORT NumLockState; if (SHORT1FROMMP(mp1)) { NumLockState = WinGetKeyState (HWND_DESKTOP, VK_NUMLOCK); Toggled = NumLockState & 0x0001; if (!Toggled) { WinSetKeyboardStateTable(HWND_DESKTOP, (PBYTE) KeyStateTable, FALSE); KeyStateTableìVK_NUMLOCK┘ |= 0x01; WinSetKeyboardStateTable(HWND_DESKTOP, (PBYTE) KeyStateTable, TRUE); } } else if (!Toggled) { KeyStateTableìVK_NUMLOCK┘ &= 0xFE; KeyStateTableìVK_CTRL┘ = 0; KeyStateTableìVK_ALT┘ = 0; WinSetKeyboardStateTable(HWND_DESKTOP, (PBYTE) KeyStateTable, TRUE); } return (WinDefDlgProc(hwnd, msg, mp1, mp2)); } } } Patrick Malneritch, Delta Air Lines ----- OS2PRG CFORUM appended at 13:16:42 on 92/07/10 GMT (by KWINZ at RALVM5) - ..... OS2PRG CFORUM modified at 13:50:39 on 92/07/10 GMT (by KWINZ at RALVM5) . Subject: Drive letter -> drive number Ref: None | Urk. Never mind. Looks like DosOpen is happy with opening | "a:panda.ndf" anyway, so I don't need to mess with the path. | There was an unrelated bug that was causing the DosOpen to | fail. Shows what I get for jumping to conclusions.... Given a drive letter, is there any way for me to find out the drive number that's required for DosQCurDir? What I need to do is to take an input of the form x:filename.ext, and insert the fully qualified path for the current directory on that drive, to come up with x:\panda\filename.ext. That should give me a valid pathname that I can pass to DosOpen. The problem is that I have a drive letter passed in, but DosQCurDir wants a drive number. Is the drive number an absolute or a relative thing? For example, if I get a drive letter "F:", will that always be drive number 6, or does it depend on how many logical drives are defined with lower letters on the system? Kim Winz ----- OS2PRG CFORUM appended at 13:33:18 on 92/07/10 GMT (by LECTION at BETASVM2) Subject: Drive letter -> drive number Ref: Append at 13:16:42 on 92/07/10 GMT (by KWINZ at RALVM5) Kim, It is my impression that the drive number is an absolute number, with drive A: starting at zero. So if you have a letter you could use the following c code to get the number: int iDriveNumber = (int) ( toupper('x') - 'A' ); Where "x" is the drive letter. David B. Lection ----- OS2PRG CFORUM appended at 13:51:54 on 92/07/10 GMT (by KWINZ at RALVM5) - Subject: Drive letter -> drive number Ref: Append at 13:33:18 on 92/07/10 GMT (by LECTION at BETASVM2) I just found out that I was jumping to conclusions and I don't need to do this after all, but thanks for the info anyway. Its good to know that; I'll file it away. Thanks, Kim Winz ----- OS2PRG CFORUM appended at 13:55:17 on 92/07/10 GMT (by 64924111 at TOROHON1) Subject: Drive letter -> drive number Ref: Append at 13:51:54 on 92/07/10 GMT (by KWINZ at RALVM5) I believe that you only need the drive number to change drives from inside a program. Regards.....Pete Liukaitis ----- OS2PRG CFORUM appended at 16:53:07 on 92/07/10 GMT (by IL12431 at OS2CUST) Subject: Changing Keyboard States Ref: Append at 14:21:19 on 92/07/09 GMT (by DALNK03 at OS2CUST) Thanks very much for the fragment, Patrick. Dave Scheffer, ARGO Data Resource Corp. ----- OS2PRG CFORUM appended at 18:23:03 on 92/07/10 GMT (by IL22868 at OS2CUST) Subject: Named Pipe Local Host Name? Ref: Append at 17:17:40 on 92/06/29 GMT (by WZ00576 at OS2CUST) I believe the WHOAMI program in the Software Library section of this BBS has what you need (it contains both C source and EXE). Ken Singer, Shell Oil, Houston ----- OS2PRG CFORUM appended at 18:48:25 on 92/07/10 GMT (by IL22868 at OS2CUST) Subject: API For PSTAT.EXE Information Ref: Append at 18:03:29 on 92/06/25 GMT (by AT at VIA) ----- I have info on DosQProcStatus for 2.0 -- got it from KILLM2 on Gilmore's Magnum BBS (it may also be on other OS/2 BBSs). If you can't find it, I may be able to FAX you the source. Also on Gilmore's BBS, but in a more restrictive area, is PROCTST. KILLM2 includes source and an include file for the structure of the buffer returned from the OS/2 2.0 version of DosQProcStatus. Following is the README.DOC file for KILLM2: =============================================================================== KILLEM.EXE kills running processes by name. It works with version 6.177h or OS/2 2.0. Haven't tried any version before or after. It uses the undocumented API DosQProcStatus to get a list of running processes - therefore its stability is in question because IBM could change or remove this API without telling anyone. Usage: KillEm process-name process-name ... If no extension is given for process-name, .EXE is appended. process-name is the name that the program was started with. All instances of all process-names will be killed unless those processes ignore KillProcess signals. Any comments, suggestions, bug reports can be sent to my CIS id 72251,750. Hope this utility proves useful. Rick Fishman Code Blazers, Inc. 4113 Apricot Irvine, CA 92720 =============================================================================== Ken Singer, Shell Oil, Houston ----- OS2PRG CFORUM appended at 12:09:13 on 92/07/13 GMT (by IL62623 at OS2CUST) Subject: Named Pipe Local Host Name? Ref: Append at 18:23:03 on 92/07/10 GMT (by IL22868 at OS2CUST) Too bad IBM still hasn't supplied me with a version of PMHPO that can download. I'll ask my SE to fetch it. Thanks. ----- OS2PRG CFORUM appended at 18:50:08 on 92/07/14 GMT (by KURODA at RALVM5) Subject : DosReallocSeg return code I get return codes ERROR_ACCESS_DENIED (5) and ERROR_LOCKED (212) from DosReallocSeg. I want to know what the return codes means. Does anyone help me ? Thanks in advance, Aki Kuroda ----- Yamato, Japan ----- OS2PRG CFORUM appended at 00:41:32 on 92/07/15 GMT (by PSC31 at OS2CUST) Subject: 32-bit executable file format Is the 32-bit executable file format documented. If so, where? Bob Dedrick, Prodigy Services Company (P*HFDD60A) ----- OS2PRG CFORUM appended at 00:52:53 on 92/07/15 GMT (by KENKAHN at PKSMRVM) Subject: DosReallocSeg return code Ref: Append at 18:50:08 on 92/07/14 GMT (by KURODA at RALVM5) There was a lot of discussion on this RC=5 problem with DosReallocSeg (I was getting it myself). I'll see if I can dig up the exact discussion but I do remember that the *fix* was to simply reissue the DosReallocSeg call if you got a RC=5 on the first try. The second try usually works. Kenneth Kahn Staff of IBM Fellow/NetWork Computing Poughkeepsie, NY Internet: KENKAHN@PKSMRVM.VNET.IBM.COM IBMMAIL: USIB25P9 at IBMMAIL ----- OS2PRG CFORUM appended at 01:36:03 on 92/07/15 GMT (by V$IPAULD at BCRVMPC1) Subject: 32-bit executable file format Ref: Append at 00:41:32 on 92/07/15 GMT (by PSC31 at OS2CUST) The header files are available as part of the toolkit, along with a program EXEHDR that dumps info from an executable. Documentation on the format has been made available external to IBM. I can print and mail you a copy, or mail you a diskette, or transfer with a modem. Get me a mailing address or a phone number ... Paul Devriendt, OS/2 Development. ----- OS2PRG CFORUM appended at 12:37:04 on 92/07/15 GMT (by KURODA at RALVM5) Subject : DosReallocSeg return code Ref: Append at 00:52:53 on 92/07/15 GMT (by KENKAHN at PKSMRVM) Kenneth, thanks for your response, I call DosReallocSeg in my own memory management component. This component calls RodSubSet, DosSubAlloc, and DosSubFree to manage the memory. This component is a set of functions like library and the other components(thread) share this component to manage the memory pool. I'm worried that I made some bug in my code. Please let me know if you find some more information. Thanks again, Aki Kuroda ----- Yamato, Japan ----- OS2PRG CFORUM appended at 23:20:39 on 92/07/15 GMT (by PSC31 at OS2CUST) Subject: 32-bit executable file format Ref: Append at 01:36:03 on 92/07/15 GMT (by V$IPAULD at BCRVMPC1) Paul, Probably U.S. mail is easiest. My address is Bob Dedrick Systems Development Manager Prodigy Services Company 445 Hamilton Avenue, Mail Stop B12D White Plains, NY 10601 We wrote a 16-bit post-mortem debugger for 1.3 which we're porting to 2.0. Thanks. Bob Dedrick, Prodigy Services Company (P*HFDD60A) ----- OS2PRG CFORUM appended at 14:42:49 on 92/07/16 GMT (by IL61500 at OS2CUST) Subject: SYTOS PLUS FOR OS/2 HELLO EVERYONE!!! DOES ANYONE KNOW OF A WAY THIS SCENARIO CAN BE ACCOMPLISHED::: USING SYTOS PLUS FOR OS/2 1.3 ON AN OS/2 LANSERVER 1.3 TO B/U TO A SCSI ATTACHE D TECMAR B/U UNIT. UNIT IS ATTACHED TO SERVER; SW IS INSTALLED AT SERVER. WOULD LIKE TO WRITE A PROGRAM THAT WOULD ACCESS SYTOS PLUS, CHECK THE HEADER ON THE CURRENT INSERTED TAPE AND RETURN A CONDITION TO ADVISE A USER TO CHANGE TH E TAPE IF IT MATCHES THE ONE OBSERVED PREVIOUSLY. SYTRON HAS BEEN OF NO ASSISTA NCE. I KNOW THIS IS A TOUGH ONE. ANY IDEAS? THANKS!! LOU LOU BREIT, DEAN WITTER REYNOLDS ----- OS2PRG CFORUM appended at 16:05:54 on 92/07/16 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? I just took a program that was working when entirely linked as a single module and broke it into a small executable and two dlls. Now, when I run it (it is a pmtype:vio) from command line, it exits immediately. From ipmd, I learn that it terminates someplace in the startup code. I don't know where to look next. Can someone give me a hint? ----- OS2PRG CFORUM appended at 16:22:51 on 92/07/16 GMT (by KARRI at WINVMK) - Subject: Exit from executable entry, what do I do? Ref: Append at 16:05:54 on 92/07/16 GMT (by IL62623 at OS2CUST) Hello, Did you write your own DLL initialisation routine or are you letting C Set/2 handle it for you? Cheers, Karri. ----- OS2PRG CFORUM appended at 17:07:05 on 92/07/16 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 16:22:51 on 92/07/16 GMT (by KARRI at WINVMK) - I have no custom init routines. My .def file looks like: LIBRARY libos PROTMODE DATA MULTIPLE NONSHARED READWRITE LOADONCALL CODE LOADONCALL EXPORTS and then the exports list. I'm using tcpip 1.2.1, which is 16 bit. Is it possible that PROTMODE is a bad idea if my dll calls into the TCP dll which is 16 bit? ----- OS2PRG CFORUM appended at 17:07:36 on 92/07/16 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 16:22:51 on 92/07/16 GMT (by KARRI at WINVMK) - (no text entered) ----- OS2PRG CFORUM appended at 08:27:04 on 92/07/17 GMT (by KARRI at WINVMK) - Subject: Exit from executable entry, what do I do? Ref: Append at 17:07:05 on 92/07/16 GMT (by IL62623 at OS2CUST) I suppose the 1st ? I have to ask you is that: is your DLL 32-bit? i.e. for OS/2 2.00? Would help a lot if you could either append a code snippet or even send it over so that I could have a look and a play around with it. All the Best, Karri. ----- OS2PRG CFORUM appended at 11:17:19 on 92/07/17 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 08:27:04 on 92/07/17 GMT (by KARRI at WINVMK) - It is a brand new 32 bit dll, built with icc and link386. I don't have any way to upload it at this time, except perhaps with help from my SE. It would be moderately large. Can I fedex you a disk? ----- OS2PRG CFORUM appended at 11:33:43 on 92/07/17 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 11:17:19 on 92/07/17 GMT (by IL62623 at OS2CUST) Some more experiments show that the dll's are themselves fine -- I can create a trivial executable that calls them, link it with them, and it works. So there is something wrong with the executable I'm actually trying to run. Is there any way to really verify that I haven't accidently compiled a .obj file with the /Ge- flag and then tried to link it into an executable? The failing program is pretty large and complex, and worked when the whole mess was compiled /Ge+ and linked all into one .exe. Surely there are a limited number of conditions that would provoke this sort of behavior? ----- OS2PRG CFORUM appended at 11:47:55 on 92/07/17 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 11:33:43 on 92/07/17 GMT (by IL62623 at OS2CUST) Well, indeed, one module linked to my base executable had been mistakenly compiled /Ge-. Couldn't link386 somehow detect and complain about this? ----- OS2PRG CFORUM appended at 13:12:34 on 92/07/17 GMT (by KARRI at WINVMK) - Subject: Exit from executable entry, what do I do? Ref: Append at 11:47:55 on 92/07/17 GMT (by IL62623 at OS2CUST) Did it fix the problem though? Cheers, Karri. ----- OS2PRG CFORUM appended at 19:02:07 on 92/07/17 GMT (by IL62623 at OS2CUST) Subject: Exit from executable entry, what do I do? Ref: Append at 13:12:34 on 92/07/17 GMT (by KARRI at WINVMK) - Oh yes. Once I compiled all the executable pieces /Ge+ the thing worked fine. ----- OS2PRG CFORUM appended at 19:08:13 on 92/07/17 GMT (by IL62623 at OS2CUST) Subject: longjmp versus exception handling Ive an exception handler. The exception handler calls a routine that calls longjmp, which of course call Unwind. The unwind target is scoped within the exception handler, so the unwind operation should not disturb the exception handler. However, it appears that once the unwind happens, the exception handler is somehow disabled, even though it still is listed in the chain rooted in the TIB. That is, a general protection fault happening after the longjmp is not reported to the exception handler. help? ----- OS2PRG CFORUM appended at 21:02:48 on 92/07/29 GMT (by IL71794 at OS2CUST) Subject: TRAP screen information? Can anyone direct me to or give me information on interpreting TRAP screens? Chris Eytalis ----- OS2PRG CFORUM appended at 15:56:40 on 92/07/30 GMT (by DALNK03 at OS2CUST) Subject: TRAP screen information? Ref: Append at 21:02:48 on 92/07/29 GMT (by IL71794 at OS2CUST) I put together this info from what I've been able to figure out. Hope it helps: To make use of the Trap Screen Information: 1 Add /Fc to C compiler options 2 Add /MAP to Link options 3 Look at the IP number from the Trap Screen 4 Search the map file for the routine that the number falls into. (the bottom information in the map file is sorted by address order) Example: IP=4194 from trap screen. Map file listings: . . . 0001:33B6 _print_ifdef_script 0001:3840 _find_and_hash_each_line 0001:443E _read_files 0001:462A _print_normal_script . . . The trap happened in routine _find_and_hash_each_line (IP > 3840 and IP < 443E 5 Subtract the offending functions address from the IP 4194 - 3840 = 960 hex. 6 Now search the .COD file which was generated (/Fc option) for the function (_find_and_hash_each_line) example: _find_and_hash_each_line PROC FAR *** 000278 c8 1a 00 00 enter 26,0 7 Take the hex number (000278) and add it to the calculated number from step 5. (278+960 = BD8 hex) 8 Now search down the .COD file for the number calculated in step 7. The trap happened at that instruction. example: ;|*** if (c != *--p1) ; Line 442 *** 000bd2 8e 46 e2 mov es,WORD PTR ìbp-30┘ *** 000bd5 49 dec cx *** 000bd6 8b d9 mov bx,cx trapped here -> *** 000bd8 26 38 07 cmp BYTE PTR es:ìbx┘,al *** 000bdb 75 12 jne $L20084 The bug is on line 442 of the source code. The CS value seems to be translatable to the .EXE's .MAP file via the following assignments: 0047=1:, 0067=2:, 0087=3:, 00a7=4:, ... Patrick Malneritch, Delta Air Lines ----- OS2PRG CFORUM appended at 18:23:47 on 92/07/30 GMT (by WEEKS at AUSVM2) - Subject: TRAP screen information? Ref: Append at 15:56:40 on 92/07/30 GMT (by DALNK03 at OS2CUST) This is a good technique for OS/2 1.x. But it is not applicable to 2.0 where IP is really EIP and is 32-bits long. I haven't figured out how to map the EIP value back to the .EXE or .DLL, much less the actual routine that was running. I'm sure our customers (and even we internal types) would love to see some guidance on how to do this in the 2.0 environment. Post it here if you have any. Craig Weeks LAN Systems Performance (DCE) Austin, Texas ----- OS2PRG CFORUM appended at 14:14:22 on 92/07/31 GMT (by KEENLEY at TOROLAB6) Subject: TRAP screen information? Ref: Append at 18:23:47 on 92/07/30 GMT (by WEEKS at AUSVM2) EXEs are always loaded at address 0x10000 (64K). DLLs are always loaded in high memory ie. somewhere below 512M. If you get a trap at a relatively low address then the trap is probably occurring inside the EXE. Subtract 0x10000 from the EIP value and look up this offset in the MAP file. This will give you the function where the trap occurred. Finding TRAPs in DLLs is harder because you have to figure out where they are loaded. John Keenleyside OS/2 Compiler Dev. ----- OS2PRG CFORUM appended at 17:41:53 on 92/08/01 GMT (by XXPRMN01 at TORVMCOP) Subject: Trap information One way of interpreting the trap info is to take the address of your first function within one of your DLL's. Usually, you will be able to then compute the address similar to the method used to find the modules within the .EXE by subtracting 0x10000. Matt Smith, Prominare Inc. ----- OS2PRG CFORUM appended at 14:14:30 on 92/08/04 GMT (by IL12431 at OS2CUST) Subject: Changing Keyboard States Ref: Append at 14:21:19 on 92/07/09 GMT (by DALNK03 at OS2CUST) I note that this fragment is a WindowProc - will it work for a VIO/Full-screen session as well? If not, does someone have a fragment for non-window sessions? Dave Scheffer, ARGO Data Resource Corp. ----- OS2PRG CFORUM appended at 15:41:05 on 92/08/04 GMT (by DALNK03 at OS2CUST) Subject: Changing Keyboard States Ref: Append at 14:14:30 on 92/08/04 GMT (by IL12431 at OS2CUST) I think the code should work for a VIO program. The WinGetKeyState function requires a message queue, so you'll need to have the following: hab = WinInitialize (0); hmq = WinCreateMsgQueue (hab, 0); I don't know how a VIO program can determine if it is gaining or losing focus, maybe somebody else can help out there. Pat Malneritch, Delta Air Lines ----- OS2PRG CFORUM appended at 18:46:34 on 92/08/05 GMT (by AMRT071 at OS2CUST) Subject: DosQueue and memory problem Are there any known problems using Dos queues ?? I have had a random problem of memory being freed during DosReadQueue and DosWriteQueue function calls. My app is multi-threaded and reads/writes to several queues. One thread creates the queues and seperate threads read from them. The problem is as follows : 1. A sperate process running on the workstation will allocate a block of memory and send it back to my app via DDE. 2. One thread will allocate a small block of memory that contains the data I wish to pass to the other thread and writes that pointer addess to a queue. 3. A seperate thread will read the pointer address from that queue and the block memory that was allocated by the seperate process is freed in error by the Que function and the value at the pointer address will be put in its place. The memory coming from the seperate process IS IN NO WAY used during or at anytime near the time the queue functions are being processed. I really need help with this one. Thanks W. David Ashley American Airlines ----- OS2PRG CFORUM appended at 23:48:00 on 92/08/05 GMT (by SDETWEIL at DETVMIC4) Subject: DosQueue and memory problem Ref: Append at 18:46:34 on 92/08/05 GMT (by AMRT071 at OS2CUST) Queues AND DDE do NOT copy data. They pass only POINTERS to data. If on the DDE send from app2 to app1, the PM WinProc should malloc() and COPY the data sent from app2, OR DosWriteQueue with the address, or it will be lost after the Winproc ends. DosReadQueue does NOT copy data. It returns the POINTER VALUE written at the time of DosWriteQueue. I've used Queues extensively and have never encountered this problem. Is is possible that the shared memeory allocated in app2 (it MUST BE SHARED for DDE between processes) is not DosGiveSeg'd to app1? Sam ----- OS2PRG CFORUM appended at 17:58:29 on 92/08/06 GMT (by IL57044 at OS2CUST) Subject: Trap D in DosSubAlloc I Have an app that uses DosAllocSeg and DosSubAlloc to obtain memory. After many iterations of DosSubAlloc/DosSubFree I am getting a trap D. Message SYS31 75 indicates DOSCALL1.DLL 0005 as the routine getting the access violation. Also, when I request 64K of memory on DosAllocSeg, OS/2 intermittantly gets a system trap D. This is an OS/2 16 bit app developed on OS/2 1.3. Is there something I should be aware of when useing this type of memory management on OS/2 2.0? Thanks! Larry Snyders ----- OS2PRG CFORUM appended at 19:34:42 on 92/08/11 GMT (by JAMILLE1 at CHGVMIC1) Subject: WinQuerySwitchList A novice C and OS/2 API programmer needs some help, please! I have written a program that looks at the task list entries and selects one to become the active program. Excerpts of the code follows: typedef struct _SWB { ULONG cswentry; SWENTRY aswentry<48>; /* square brackets for an array */ } SWB; /* I am establishing a max of 48 entries */ HAB hab = 0; SWB SwitchBlock; void main () { ULONG NumberOfTasks,i; NumberOfTasks = WinQuerySwitchList(hab,NULL,0); NumberOfTasks = WinquerySwitchList( (HAB) hab, (PSWBLOCK) &SwitchBlock, (ULONG) sizeof(SwitchBlock) ); etc. My problem is that it keeps blowing up on the last statement with a "General Protection Fault Exception". (I step it through debug and the 1st WinQuerySwitchList runs fine and returns a valid number in NumberOfTasks.) I'm doing this with OS/2 2.0 and C-SET/2. Any help would be appreciated. Thanks. ----- OS2PRG CFORUM appended at 11:22:01 on 92/08/13 GMT (by DALNK03 at OS2CUST) ..... OS2PRG CFORUM modified at 11:32:07 on 92/08/13 GMT (by DALNK03 at OS2CUST) Subject: WinQuerySwitchList This APPEND was deleted at 07:32:07 on 08/13/92 EST. ----- OS2PRG CFORUM appended at 19:31:02 on 92/08/17 GMT (by OCANUC at BETASVM2) Subject: Trap D in DosSubAlloc Ref: Append at 17:58:29 on 92/08/06 GMT (by IL57044 at OS2CUST) DosAllocSeg no longer exists in OS/2 2.0 due to the flat memory model. As for memory sub-allocation the internal discussion within IBM seems to indicate that the malloc and free support in the C-Set/2 product are just fine for most applications (we use this in particular). Brandon Booth Spoken Language Systems ----- OS2PRG CFORUM appended at 19:50:39 on 92/08/17 GMT (by FIECHTNE at BCRVMPC1) Subject: Trap D in DosSubAlloc Ref: Append at 19:31:02 on 92/08/17 GMT (by OCANUC at BETASVM2) DosAllocSeg certainly DOES exist in 2.0 but not in the 32-bit API set. Please re-read the append which you referenced, this is a compatability point which we MUST maintain. As for solving the original problem, I've not seen this type of problem and I have a home grown 16-bit Async Terminal which uses these calls all over the place. I know of no restrictions that are imposed by OS/2 2.0 for these services. I suggest a service call to log a problem report against our 1.X compatability. Mark Fiechtner (OS/2 Development) ----- OS2PRG CFORUM appended at 23:05:17 on 92/08/17 GMT (by IL72906 at OS2CUST) Subject: Writing a IFS? Where is there information on writing an Installable File System? David Alcock - Central and South West Services - Dallas, Tx ----- OS2PRG CFORUM appended at 12:47:50 on 92/08/18 GMT (by 72468326 at EHONE) Subject: OS/2 V2 and Fortran/2 Malfuntions with READ from a file My customer uses large programs compiled with Fortran/2. If he runs such programs under OS/2 V2 he gets errors when reading real variables from a file. For example: READ (5,'(F10.3)'),A ( A = 1.000) WRITE (7,'(F10,3)'),A ( A = +++++++++++) Who has any ideas, suggestion? Please respond offline to Wilfried Mensching, System Engineer ACIS Hannover, Germany HAMENSC at HAMBVM2 ----- OS2PRG CFORUM appended at 19:49:19 on 92/08/20 GMT (by IL12431 at OS2CUST) Subject: OS2 v1.3 VIO 132 column mode What VIO function mode is used to get 132 column display? The DOS mode 20 was mode 20, and it is not clear to me in the documentation how to toggle 132 column VIO setmode. Dave Scheffer, ARGO Data Resource Corp. ----- OS2PRG CFORUM appended at 20:07:06 on 92/08/20 GMT (by KENKAHN at PKSMRVM) Subject: OS2 v1.3 VIO 132 column mode Ref: Append at 19:49:19 on 92/08/20 GMT (by IL12431 at OS2CUST) You use the VioSetMode command. To swtich to 132 columns you use: viomi_Col = 132; viomi_Row = 28; viomi_hres = 1056; VioSetMode(..); To switch back to 80 column mode viomi_Col = 80; viomi_Row = 25; viomi_hres = 720; VioSetMode(..) The important part here is setting the Horizontal resolution. Kenneth Kahn Staff of IBM Fellow/NetWork Computing Poughkeepsie, NY Internet: KENKAHN@PKSMRVM.VNET.IBM.COM IBMMAIL: USIB25P9 at IBMMAIL ----- OS2PRG CFORUM appended at 20:36:35 on 92/08/21 GMT (by IL12431 at OS2CUST) Subject: OS2 v1.3 VIO 132 column mode Ref: Append at 20:07:06 on 92/08/20 GMT (by KENKAHN at PKSMRVM) Ken, thanks for the info. However, when I run this code under OS/2 2.0 the result is a shaking, banded display and appears to be scrolling very rapidly from top to bottom. Following are the parameters on the call. Length = 34 Type = 1 Color = 4 Col = 132 Row = 28 hres = 1056 vres = 400 Format ID = 0 Attrib = 1 Interestingly enough, I get similar results on the MODE 132,28 command. I do get a zero return code from the system when I pipe my test code through CVP, though. Could the problem be hardware? I'm testing on a beta model 57 w/8515 monitor. The BIOS is dated 8/28/91, submodel 38, revision 6. ----- OS2PRG CFORUM appended at 00:07:53 on 92/08/22 GMT (by KENKAHN at PKSMRVM) Subject: OS2 v1.3 VIO 132 column mode Ref: Append at 20:36:35 on 92/08/21 GMT (by IL12431 at OS2CUST) I'm fairly certain it's not an OS/2 V2 problem per say. I got this code from an application I have (a 3270 Terminal Emulator) that runs on both 1.x and 2.0 systems. I'm afraid I'm not that savy on this so hopefully someone else can comment on your problem. Just to make sure I got the obvious out of the way, you are using an XGA type adapter, right? 132 columns are supported only on an XGA. Also this will olny work in a full screen mode, not in a VIO window. Kenneth Kahn Staff of IBM Fellow/NetWork Computing Poughkeepsie, NY Internet: KENKAHN@PKSMRVM.VNET.IBM.COM IBMMAIL: USIB25P9 at IBMMAIL ----- OS2PRG CFORUM appended at 20:05:40 on 92/08/24 GMT (by DJJQ1 at OS2CUST) Subject: Writing a IFS? Ref: Append at 23:05:17 on 92/08/17 GMT (by IL72906 at OS2CUST) I don't have a document ID, but we obtained from IBM a document titled "Installable File Systems For OS/2 Version 2.0" from the OS/2 File Systems Department, PSPC Boca Raton, Florida (dated Feb 17, 1992) It doesn't have "IBM Internal use only" on it so I assume that you should be able to get it without too much trouble. I'm not personally working with IFS, but I understand that it is not trivial. Hope this helps Kevin Phillips - DST Systems, Inc. Kansas City, MO ----- OS2PRG CFORUM appended at 18:01:38 on 92/08/25 GMT (by AMRT071 at OS2CUST) Subject: Pgms using IOPL I am currently writing a program which needs to play some music to the PC's speaker. All the examples I am using are 1.x programs which use segments which have been marked IOPL to the linker. I have tried using the same technique with 32-bit programs but when executed the segments which have been marked as IOPL all cause an access violation when the first code instruction (a PUSH EBP) is executed. Can someone give me a hint as to what I am doing wrong? Thanks in advance. W. David Ashley American Airlines ----- OS2PRG CFORUM appended at 23:52:48 on 92/08/26 GMT (by SDETWEIL at DETVMIC4) Subject: Writing a IFS? Ref: Append at 20:05:40 on 92/08/24 GMT (by DJJQ1 at OS2CUST) The hardest part is memory mgmt, as the IFS runs at ring 0 (kernel mode).. I have written a couple of IFSs supporting both 1.3 and 2.0, so ask away.. Sam ----- OS2PRG CFORUM appended at 23:54:30 on 92/08/26 GMT (by SDETWEIL at DETVMIC4) Subject: Pgms using IOPL Ref: Append at 18:01:38 on 92/08/25 GMT (by AMRT071 at OS2CUST) You still have to do the same thing, 32 bit pgms do not yet support IOPL level segments. So, you must make a 16 bit segment, mark it IOPL as before and call it from the 32 bit code.. 32 bit code CANNOT do IN/OUT instructions directly... Sam ----- OS2PRG CFORUM appended at 18:22:09 on 92/09/01 GMT (by PNWB016 at HONE81) Subject: Exception handler questions One of the developers I work with asked that I submit the following questions... thanks in advance for any light that may be shed on this. I need some help with an Exception Handler: I want to get the same information that is displayed when a trap screen appears. I have found where the EIP is kept (ExceptionAddress field of the EXCEPTIONREPORTRECORD) but... Where is the thread number kept? Where is the Register info. kept? It doesn't appear that it's in the CONTEXTRECORD field because all all values are zero in it when I get control. Thanks again! Wayne "Not responsible for the contents of this message" Holmes U S WEST Communications ----- OS2PRG CFORUM appended at 15:59:22 on 92/09/09 GMT (by STINCHI at ROMESC) Subject: DLL. We are working with OS/2 DLL. We need to export functions and variables from main program (staticly linked) to DLL (all the examples show how to export functions and variables from DLL to main). Is it possible ? Have someone source code or some form of documentation on this problem? Any help will be appreciated. Stefano Stinchi and Fabrizio Pacelli. ----- OS2PRG CFORUM appended at 16:48:52 on 92/09/09 GMT (by WEEKS at AUSVM2) - Subject: DLL. Ref: Append at 15:59:22 on 92/09/09 GMT (by STINCHI at ROMESC) You can't EXPORT from .EXE to .DLL the way you can from .DLL to .EXE (via .DEF file). If you want the .DLL to have access to the .EXE variables you can use shared memory, or pass a pointer to the .DLL function as a parameter, or any one of several other (probably) techniques. Craig Weeks LAN Systems Performance (DCE) Austin, Texas ----- OS2PRG CFORUM appended at 16:55:59 on 92/09/09 GMT (by VENRTC51 at DFWVM04) Subject: int 3 obj I need an object file that does an interrupt 3 which I can link with my C code. This is for 16 bit code, large model. Thanks, Kurt (VENRTC51 at DFWVM04) ----- OS2PRG CFORUM appended at 01:35:04 on 92/09/10 GMT (by SDETWEIL at DETVMIC4) Subject: int 3 obj Ref: Append at 16:55:59 on 92/09/09 GMT (by VENRTC51 at DFWVM04) If using MS C6, just make one trace() { _asm { int 3 }; } Then when you need to stop, just add trace(); to your code. I pass a parm, when I need to know WHICH trace entry caused the stop Sam ----- OS2PRG CFORUM appended at 20:28:46 on 92/09/14 GMT (by XXWESB07 at TORVMCOP) Subject: DosWriteQueue returns 334 I am developing a 32-bit multithreaded PM application that is using queues (ie. DosCreateQueue, DosWriteQueue, DosReadQueue). My app successfully writes/reads from/to the queue a number of times. Then a DosWriteQueue() call fails with error 334. Looking this up it says "Inadequate queue memory". Doing a help 334 it goes on to say that you should contact the supplier of the program! Anyways, I am wondering if there is a system configuration parameter that affects the amount of memory available to the queues code in the system. I am unable to locate any additional information on this error code and how to resolve it. My swap file is about 10MB and there is about 35MB left on the swap device. I am running 2.0 GA on a P70 with 8MB of RAM using the Workset/2 (no CSDs) over a network. The code I am using was copied (almost a clone) from the 16-bit version of our application that was a background process under OS/2 1.3. We never encountered this error under OS/2 1.3. Also, if the DosWriteQueue() fails, the app sleeps for 5 seconds and then retries but is never able to successfully complete. Thanks... Craig Knelsen, ISM Corp. One Research Drive, Regina, SK S4S 7H1 (306) 781-5191 ----- OS2PRG CFORUM appended at 01:36:49 on 92/09/15 GMT (by SDETWEIL at DETVMIC4) Subject: DosWriteQueue returns 334 Ref: Append at 20:28:46 on 92/09/14 GMT (by XXWESB07 at TORVMCOP) When writing between processes, one must give/get access to the shared memory somehow. Under 1.x you did a DosGive/GetSeg on the selector and then a DosFreeSeg when you were done. Both sides freed the selector eventually, and the use count dropped to 0, and the memory was freed... Now, under 2.0, the BASE address of the DosAllocMem object must be Given/Gotten AND FREED, else the memory does not get freed.. 334 sounds like the elements were not pulled off the queue, but only PEEKed... Sam ----- OS2PRG CFORUM appended at 15:54:32 on 92/09/15 GMT (by XXWESB07 at TORVMCOP) Subject: DosWriteQueue returns 334 Ref: Append at 01:36:49 on 92/09/15 GMT (by SDETWEIL at DETVMIC4) Thanks for the reply. Your reply tells me that I need to further clarify what my app is doing. The 16-bit version communicates with other processes (the one who started it and those it starts) using pipes (unnamed and named). In the 32-bit version, the process that used to start the 16-bit app is now combined into one program with multiple threads although it still communicates with 32-bit PM apps, that it starts, using named pipes. The queues are used internally within the app (either 16 or 32 bit versions) to send messages between the threads. We use malloc() to allocate space for the messages. However, since our message areas are of a fixed size, we optimize the allocation routine by saving the addresses of our processed messages in a free list. The allocation routine checks this free list and uses a previously allocated message area if there's one available. Otherwise, it calls malloc(). I hope that the system queues code in 2.0 is not expecting the msgs that have been written to the queue and then retrieved by DosReadQueue() are free'd within a short space of time after that. I do not see why it would matter in the above scenario. Craig Knelsen, ISM Corp. One Research Drive, Regina, SK S4S 7H1 (306) 781-5191 ----- OS2PRG CFORUM appended at 16:26:26 on 92/09/15 GMT (by ALIU at BOSTON) -- Subject: 'Dir' Command locks resources? Hi, I am wondering about what's going on with the simple old 'Dir' command. Basically, I'm writing a REXX program that is checking the last file write access timestamp of a given file. Everything seems OK with the REXX program itself. However, in another OS2 Window, I've got another program that will occasionally update the same file. When that happens, it appears that as I obtain the Directory information of that file via that Rexx program, if I happen to be updating that file, I get a Resource/Lock message with an RC = 32. I think I can get around this by trapping for RC = 32 in the other process basically by doing a retry of the write. However, I wonder what's going on with the 'Dir' statement. It seems to me that I'm only trying to read the Directory Information of the file. Why am I locking up the Resource? Is there some other way of getting the Directory Information without locking that resource up? Anybody else have another suggestion than checking for RC = 32? Regards, Art Liu (ALIU at BOSTON) IBM Services Company Client/Server Practice ----- OS2PRG CFORUM appended at 18:11:52 on 92/09/15 GMT (by RICHKULP at CARVM3) Subject: 'Dir' Command locks resources? Ref: Append at 16:26:26 on 92/09/15 GMT (by ALIU at BOSTON) In your REXX command try doing this command instead of 'Dir' It should be easier and faster and may fix your lock problem: filename = 'c:\whatever' /* Any file name is valid. */ . . . datetime = stream(filename, 'c', 'query datetime') datetime will then contain the date and time in this format: mm-dd-yy hh:mm:ss Rich Kulp ----- OS2PRG CFORUM appended at 03:19:54 on 92/09/16 GMT (by SDETWEIL at DETVMIC4) Subject: DosWriteQueue returns 334 Ref: Append at 15:54:32 on 92/09/15 GMT (by XXWESB07 at TORVMCOP) As the queue doesn't actually copy any data, there is no timing relationship involved. all the queue does is pass the pointer from one place to another.. I'd say, if you can recreate this in a smaller testcase, then you have a legitimate bug.. Just for grins, have you tried serializing around reading/writing from/to the queue from the multiple threads? It shouldn't be required, but then bugs are just that.. Sam ----- OS2PRG CFORUM appended at 20:02:25 on 92/09/17 GMT (by XXTCAN06 at TORVMCOP) Subject: Interface Smalltalk/V PM 1.4 to NETAPI.DLL under OS/2 2.0. Hi! I'm trying to call my 1st API in NETAPI.DLL from Smalltalk/V PM 1.4 but I'm obviously not doing it quite right yet 'cause I'm getting protection exceptions or infinite message loops et al. Firstly: I'm under OS/2 2.0 but I'm reading the usoft 'LAN Manager Programmer's Reference for version 2.0'. Is that reference book valid? If so, why would this wrapper method for NetUseEnum not work?: useEnum: pszServer level: sLevel buffer: pbBuffer bufferSize: cbBuffer entriesRead: pcEntriesRead totalAvail: pcTotalAvail <api: NETUSEENUM struct short struct ushort struct struct ushort > `self invalidArgument . The code to test this method is as follows: | answer buffer entriesRead totalAvail | buffer := String new: 64 . entriesRead := 0 . totalAvail := 0 . answer := NetAPIDLL open useEnum: '' asParameter level: 1 buffer: buffer asParameter bufferSize: (buffer size) entriesRead: entriesRead totalAvail: totalAvail . self halt . The 'NetAPIDLL open' seems to be working 'cause it inspects as 'NetAPIDLL{61 3 }'. The prototype for NetUseEnum in the book is: #define INCL_NETUSE #define INCL_NETERRORS #include <lan.h> API_FUNCTION NetUseEnum (const char far * pszServer, short sLevel, char far * pbBuffer, unsigned short cbBuffer, unsigned short far * pcEntriesRead, unsigned short far * pcTotalAvail ); Anyone with any ideas? Maybe especially Larry Mason of Texas Instruments who said, back in July, that he had some expertise in that field. Thanks. ----- OS2PRG CFORUM appended at 07:32:17 on 92/09/23 GMT (by DEV0144 at OS2CUST) Subject: ACCESSING PHYSICAL MEM VIA IOCtl Device Driver Function Howdy, I am writing a device driver, for a memory mapped i/o board. This memory mapped i/o board is an isa 8 bit isa class card that runs in an IBM pc 'crone'. I would appreciate a code sample that illustrates an application call via DosDevIOCtl to a device driver accessing a chunk of physical memory. Also the receiving device driver code chunk would be appreciated. I am thouroughly confused as to how to accomplish this seemingly simple task. Every time I access my device driver I get a 'GP' fault. I have ordered the Control Programming Reference manual, but have not received it as yet. I hope this will shed some further light on the subject. The Physical Device Driver Reference is not very clear on this subject. I appreciate IBM's decision to allow folks (such as I) access to this forum. Thankyou. Any help regarding this matter will be greatly appreciated. ----- OS2PRG CFORUM appended at 10:19:00 on 92/09/23 GMT (by KENKAHN at PKSMRVM) Subject: ACCESSING PHYSICAL MEM VIA IOCtl Device Driver Function Ref: Append at 07:32:17 on 92/09/23 GMT (by DEV0144 at OS2CUST) I thinkwhat you are asking is how an application can request access to the physical memory on an adapter card. If this is the case then the following applies (if not the following is useful anyway but you can ignore it). In the device driver, you use the PhysToUVirt DevHelp to translate the physical memory to a virtual storage address that is accessible via the calling process's LDT Table. DCA_Address DD 0CE000h ; Address of DCA Buffer on 3270 Adapter org DCAaddrHI DW 0000Ch DCAaddrLO DW 0E000h Mov AX,DCAaddrHI ; Get the High-order address Mov BX,DCAaddrLO ; Get the Low-order address Mov CX,8192 ; get the buffer size Mov DH,1 ; make segment R/W Mov DL,DevHlp_PhysToUVirt Call Device_Help Mov CX,ES ; Save Selector Mov DX,BX ; Save Offset LES BX,Request_Packet ; restore request packet address LES BX,ES:[BX+15] ; get a pointer to IOCtlParms Mov ES:[BX],DX ; save Offset into IOCtlParms Mov ES:[BX+2],CX ; Save Selector into IOCtlParms The allocated virtual address will be returned in IOCtlParms. An application would use DosDevIOCtl to call the DD and pass it a 32-bit pointer field. The DD would return the values ES:BX in this 32-bit field, which the application could then use directly as a pointer to the buffer. IOCtlData DB 50 DUP(?) IOCtlParms DB 50 DUP(?) DCApointer DD ? @DosDevIOCtl IOCtlData,IOCtlParms,2,80h,DD_Handle Mov BX,Word Ptr IOCtlParms ; get the Offset returned Mov Word Ptr DCApointer,BX ; save it Mov BX,Word Ptr IOCtlParms+2 ; get the Selector returned Mov Word Ptr DCApointer+2,BX ; save it Kenneth Kahn Staff of IBM Fellow: IBM Personal/370 Poughkeepsie, NY Internet: KENKAHN@PKSMRVM.VNET.IBM.COM IBMMAIL: USIB25P9 at IBMMAIL ----- OS2PRG CFORUM appended at 21:06:33 on 92/10/05 GMT (by DEV2953 at OS2CUST) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 I have been having the following problem in converting my OS2 1.3 Applicaation to OS2 2.0 application. We have a hardware accelerator that resides on the PC B us and the Microchannel. It only has I/O port for communicatioon. In OS2 1.3 w e were able to talk to our hardware by enabling IOPL=YES in the config.sys f ile and enabling IOPL privileges for the whole segment that specifically commun icated with the hardware. If a similar procedure is followed on OS2 then I am getting exception error whenever I enter the segment with IOPL enabled for the segment. If IOPL is not enabled then the program enters the segment but flags an exception whenever it encounters an I/O instruction. I would like some help that can resolve my problem vis-a-vis access to our hard ware aceelerators ports . We are currently using WATCOM/386 compiler for OS2. The version of the compiler is 9.0. WATCOM is suggesting that OS2 2.0 does not allowing direct access to hardware ports. "IS IT TRUE". MY REQUEST FOR AN ANSWER IS VERY URGENT. WE HAVE A MAJOR CUSTOMER WHO HAS REQ UESTED US TO PORT OUR LIBRARIES TO OS2 2.0 My phone number is 619-452-0903 x26 and I am available 12:30 p.m to 10 p.m East er Time. ----- OS2PRG CFORUM appended at 22:09:07 on 92/10/05 GMT (by ZUBER at RALVM14) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 21:06:33 on 92/10/05 GMT (by DEV2953 at OS2CUST) IOPL still exists in V2.0, but it has to still look just like it did in V1.3. Or in other words, it can't have 386 instructions. Clyde Zuber ----- OS2PRG CFORUM appended at 02:20:00 on 92/10/06 GMT (by V$IPAULD at BCRVMPC1) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 21:06:33 on 92/10/05 GMT (by DEV2953 at OS2CUST) You still have to be at ring 2 to access ports in version 2.0. You will trap if you try to access a port from ring 3. There is no compiler support for ring 2 in the flat 32-bit world (it requires a segment register change, and the concept of changing segment registers does not exist in a flat programming model. The easiest way to do this is to do the port access from ring 2 in 16-bit code just as on versions 1.x, and write a thunk to call the 16-bit code from your 32-bit code. Paul Devriendt ----- OS2PRG CFORUM appended at 12:12:30 on 92/10/06 GMT (by KEENLEY at TOROLAB6) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 02:20:00 on 92/10/06 GMT (by V$IPAULD at BCRVMPC1) C Set/2 will supply the thunking code for the 32 to 16 call automagically. John Keenleyside OS/2 Compiler Dev. ----- OS2PRG CFORUM appended at 20:25:26 on 92/10/06 GMT (by ABLI001 at OS2CUST) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 21:06:33 on 92/10/05 GMT (by DEV2953 at OS2CUST) This is interesting. I thought 2.0 was upwardly compatible in this area. Did you get the error using the original compiled version (16bit) or did it only occur after recompiling with the 386 compiler? ->Rich Corbett; ABC Liquors, Inc., Orlando, FL (NASPA: CORBRIC7) ----- OS2PRG CFORUM appended at 21:22:55 on 92/10/06 GMT (by V$IPAULD at BCRVMPC1) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 20:25:26 on 92/10/06 GMT (by ABLI001 at OS2CUST) OS/2 2.0 is compatible as regards IOPL in that a 1.x application with both ring 2 and ring 3 segments will run unchanged on 2.0. However, you can not take the source for this application, and recompile as 32-bit. Paul Devriendt ----- OS2PRG CFORUM appended at 14:21:25 on 92/10/08 GMT (by XXPRMN01 at TORVMCOP) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 12:12:30 on 92/10/06 GMT (by KEENLEY at TOROLAB6) Automagically ?? Matt Smith, Prominare Inc. ----- OS2PRG CFORUM appended at 22:06:28 on 92/10/08 GMT (by THRUMD at SYDVM1) Subject: Direct Access of I/O ports using the IOPL priveleges in OS2 2.0 Ref: Append at 14:21:25 on 92/10/08 GMT (by XXPRMN01 at TORVMCOP) Absolugely! Dave ----- OS2PRG CFORUM appended at 01:24:32 on 92/10/09 GMT (by AAAF0CY at OS2CUST) Subject: DosStartSession I would like to start a DOS window from an OS/2 program and have the OS/2 program be able to know when the DOS pgm terminates. There are several options, but so far the most promising has been to use DosStartSession creating a child window and using the TermQ field to provide notification to the OS/2 pgm. I'm having a problem where the child window immediately terminates. i.e. the window appears with "Dos Window" in the title bar then immediately "Closing DOS Window" shows up in the title bar and the window terminates. If I create the DOS window as independent instead of a child, it works fine. The queue is being created and the create function has a return code of 0. I don't remember what happens if I start an OS/2 child window. I think it also immediately terminates, so there maybe a problem with the DosStartSession call. Has anyone had experience with this under OS/2 2.0 and can give me a helpful hint??? Thanks Jim Sauber, MN Mutual ----- OS2PRG CFORUM appended at 03:52:45 on 92/10/13 GMT (by DEV2203 at OS2CUST) Subject: PDD access via DOS MVDM using IOCTLS (INT44) Can a DOS MVDM session access an OS/2 device driver (PDD) using the INT 21 function 44H to create an IOCTL packet for send and receive. The general thought here is that you need a VDD to connect to the OS/2 PDD. The DOS program would then speak to the VDD via INT21 (44) and the VDD would pass the commands over to the PDD. The objective is to allow a DOS program to talk to the existing OS/2 device driver via INT 21 calls. Can this be done without the need for a VDD? Thanks, Phil ----- OS2PRG CFORUM appended at 11:08:49 on 92/10/13 GMT (by FIECHTNE at BCRVMPC1) Subject: PDD access via DOS MVDM using IOCTLS (INT44) Ref: Append at 03:52:45 on 92/10/13 GMT (by DEV2203 at OS2CUST) This can be done without a VDD. (the MODE command does this) Mark Fiechtner ----- OS2PRG CFORUM appended at 15:44:58 on 92/10/14 GMT (by SAMEH at YKTVMH) - Subject: General OS/2 semaphore question. When using an Event semaphore, the system keeps track of the number of times that the semaphore gets posted since the last Reset. The documentation is clear on this. However, the documentation says nothing about what happens to this post count as a result of DosWaitEventSem() or DosWaitMuxWaitSem() system calls. Does the post count get automatically decremented every time I DosWaitEventSem() on the semaphore and find it already posted? Does the post count get automatically decremented when I DosWaitMuxWaitSem() and the Event semaphore is already posted? (This is after adding the event semaphore to the MuxWait semaphore). Will the semaphore eventually get automatically 'Reset' when the post count reaches zero? Also, is there a good OS/2 2.0 reference book like the "The Waite Group's OS/2 Programmer's Reference by Asael Dror" was for OS/2 1.3? Thanks in advance for any help..... Sameh A. Fakhouri ----- OS2PRG CFORUM appended at 15:49:35 on 92/10/15 GMT (by PSC30 at OS2CUST) Subject: writing mouse device drivers I originally posted this question in BASE back in May. I got some responses from KIKO @ TOROLABS (Thanks, Kiko!) but without REAL documentation and the toolkit, I couldn't react to them. I now have BOTH, and the project was overdue then. I wrote a mouse replacement driver for 1.3 to support a graphics drawing pad (absolute positioning device). When I moved this driver to 2.0, it did not work. The several suggestions offered then did not work. Basically, after the device driver loads via CONFIG.SYS it initializes correctly(?) but there is no cursor visible on the screen at all. KDB seems to show that the correct events are being pushed to MOUSE$ via the Process_ Packet (or Process_Absolute) IDC. The Query_Config and Read_Enable entries seem to be entered correctly. If I run a text program that issues 'MouRead EventQue' it shows ever changing (and apparently correct) positions - so I assume that the events being pushed to MOUSE$ are being accepted and reported. My suspicion is that somehow the cursor (pointer arrow) "thinks" it is "behind" the desktop, but I have no way of knowing that and I have not yet been able to externalize it into visibility. Kiko offered the suggestion that ALL 2.0 mouse events MUST show motion (event bit 0 = 1), but I have tried that without success and the Toolkit PDD manual EXPLICITLY says that absolute position events do not contain motion. I also run the device in "relative" mode (Process_Packet) with no success. I know that the 2.0 IDC structures have changed slightly from 1.3. I've programmed for that! In 1.3, the AX register contained non-zero if MOUSE$ did not accept the event (the carry bit was also set per the doc, but AX was also set - I walked through the MOUSE$ code too many times!!) I am consistently getting a non-zero AX in 2.0 (documentation claims only carry is significant, but carry is off in my case so I must assume that the event was accepted in spite of the AX code... I'm starting to run on here. Can anybody help with this? I think if I can get the pointer visible, it would be working correctly.... Graham Yarbrough - Prodigy Services Co ----- OS2PRG CFORUM appended at 15:33:41 on 92/10/16 GMT (by PSC30 at OS2CUST) Subject: Writing a IFS? Ref: Append at 23:05:17 on 92/08/17 GMT (by Il72906 at OS2CUST) I downloaded a copy of "Installable File Systems for OS2 2.0 ....." dated May 20, 1992 from the Developers Assistance BB in Boca. I suggest that you start by calling 407-982-6408 to get enrolled. All you need is a license for the 2.0 Toolkit. It is a big document and it looks like writing an IFS is non-trivial, but it doesn't look like its beyond a good OS/2 knowledgeable programmer. Graham Yarbrough - Prodigy Services Co ----- OS2PRG CFORUM appended at 12:48:32 on 92/10/21 GMT (by IL84740 at OS2CUST) Subject: Programmatic startup of Windows Applications Does anyone know how to programmatically start up a Windows application in a seamless window? I want to do the equivalent of "START /WIN PROGRAM.EXE" without kicking off a command processor (CMD.EXE). I have tried starting WINOS2.COM with both DosStartSession and WinStartApp to no avail. Are there some parameters I need to send to WINOS2.COM so it only starts the application and not a full Windows session (i.e. Program Manager...)? Rick Newill Ford Motor Company ----- OS2PRG CFORUM appended at 12:56:29 on 92/10/21 GMT (by THRUMD at SYDVM1) ..... OS2PRG CFORUM modified at 13:00:37 on 92/10/21 GMT (by THRUMD at SYDVM1) Subject: Programmatic startup of Windows Applications Ref: Append at 12:48:32 on 92/10/21 GMT (by IL84740 at OS2CUST) | it's late - in retrospect I didn't answer your question. | I guess it's time to go home... Dave ----- OS2PRG CFORUM appended at 17:31:05 on 92/10/21 GMT (by AAAF0CY at OS2CUST) Subject: Programmatic startup of Windows Applications Ref: Append at 12:48:32 on 92/10/21 GMT (by IL84740 at OS2CUST) Using DosStartSession, you can pass in the environment string settings for the object and DOS settings. If you send "SETPROGTYPE=WINDOWEDWIN;" along with the other settings (including setting the DOS environment, i.e. EMS_MEMORY_LIMIT, ...), this should make it a windows windowed session. I believe that for the program name you only have to name the executable. Jim Sauber, MN Mutual ----- OS2PRG CFORUM appended at 08:03:47 on 92/10/22 GMT (by DEV4186 at OS2CUST) Subject: DosFindFirst/Next() Can't Find Leading spaces Ok, Here is an interesting one for all of you. Has anyone tried to locate Files with DosFindFirst()? Write a simple program that Call FindFirst, printf, while( anymore files?){DosFindNext. printf} Compile ICC test.c Default Compile and Link. Run The Given Program in a Directory with filenames that contain leading spaces They don't show up! I have an APAR open on this one, but thought I would see if anyone had encountered this one yet? Thanks Michael Kupka ----- OS2PRG CFORUM appended at 03:18:58 on 92/10/27 GMT (by HHSD012 at OS2CUST) Subject: Starting a DOS app in the foreground Before I embark on trying to write a C program that would start a DOS app in the foreground of the WPS, I need to know if it is even possible. Problem is that I have a company grown Windoze app that starts some DOS apps. When it does, it starts them in the foreground of Windoze so as to give the user easier access to them for response. As everybody knows, OS/2 does not allow a Windows program to start a DOS program. So we have the Windows programs write the command line for the DOS app to a file where we have a REXX program looking for writes to the file. When the file is written to, it executes the command line. Unfortunately, the user then has to see the error about not being able to start a DOS app and at that time, he or she CNTR-ESCapes to the Window List to pick out the DOS app for use. I was told that the only way to get the DOS app to start in the foreground was by the use of C programming. Now I am getting a conflicting statement from a company that does OS/2 programming for a living. They say that the only way to start a DOS app in the foreground is for the program starting it to be in the foreground. This of course is impossible to accomplish. I know that it is also false, since I can get a notification of a print job being finished over the network. Maybe it is totally different, but I think that there should be some way, after all we are talking about control of the interface (WPS), which a C programmer should have some of. Heck, even if the program which starts the DOS app has to give itself the focus before starting the DOS app, I can live with that. Any suggestions or definitive answers will be appreciated. Thanks, Nick Gardiner HHSD012 ----- OS2PRG CFORUM appended at 16:40:33 on 92/10/27 GMT (by NAAO3IY at HONE82) Subject: Starting a DOS app in the foreground Ref: Append at 03:18:58 on 92/10/27 GMT (by HHSD012 at OS2CUST) Why not just start a CMD that executes the DOS program? John Kennedy, Hoechst Celanese (The OS/2 Hobbit) ----- OS2PRG CFORUM appended at 18:54:23 on 92/10/27 GMT (by MAAV7IB at HONE83) Subject: Debugging 16-bit code under OS/2 2.0 We have an existing OS/2 1.3 application that we want to move to OS/2 2.0. Our concern is supporting (debugging) the 16-bit code under OS/2 2.0. The current organization is the Application Manager 3 (AM) programs call MS C 6.0 code, which calls Sybase "Open Client", TCP/IP, and APPC routines. We want to call this existing C code from AM 4 (on OS/2 2.0). We think we can link the existing 16-bit MS C code with AM 4, but are concerned with debugging the code under 2.0. We'd rather not convert the existing code to C-Set/2 (32-bit) for the time being. What can I use to debug the code under 2.0? Are there facilities in the Workset/2 that I can use? OS/2 itself? Or should I be looking at 3rd party packages? Any advise would help. Thanks in advance! Kevin Miller, The Capital Group, Inc. (USCAPFAB at IBMMAIL) ----- OS2PRG CFORUM appended at 18:56:52 on 92/10/27 GMT (by HHSD012 at OS2CUST) Subject: Starting a DOS app in the foreground Ref: Append at 16:40:33 on 92/10/27 GMT (by NAAO3IY at HONE82) Oh, if it were only that easy. We did start off with a REXX that was supposed to start the session in the foreground. But; when you run it and the CMD is not in the foreground (which it can't be), it gives you an error about not being able to start the session in the foreground and then starts it in the background. That is the point at which I was told I was going to have to go to C to force the session into the foreground and hence my concern about reports that even mighty C cannot do this...Thanks Nick Gardiner HHSD012 ----- OS2PRG CFORUM appended at 20:15:50 on 92/10/27 GMT (by WZ00813 at LINK21) Subject: Order of mail from a mailslot I am writing a distributed application using LAN Server as the LAN OS. I am using mailslots to send messages and data between computers. I have run into the problem of having the messages and data come from the mailslot in a different order than they were written. Is this the way mailslots are supposed to work?? These messages/data blocks are all first class/priority one. Ross D. Schaack ----- OS2PRG CFORUM appended at 21:42:56 on 92/10/27 GMT (by WEEKS at AUSVM2) - Subject: Debugging 16-bit code under OS/2 2.0 Ref: Append at 18:54:23 on 92/10/27 GMT (by MAAV7IB at HONE83) I believe IPMD (the C Set/2 debugger) can decode 16-bit code that was compiled for Codeview. Ask over in C-SET2 CFORUM to be sure. Craig Weeks LAN Systems Performance (DCE) Austin, Texas ----- OS2PRG CFORUM appended at 16:31:39 on 92/10/28 GMT (by NAAO3IY at HONE85) Subject: Starting a DOS app in the foreground Ref: Append at 18:56:52 on 92/10/27 GMT (by HHSD012 at OS2CUST) Try using a C program to start a CMD in the foreground that executes the DOS program. John Kennedy, Hoechst Celanese (The OS/2 Hobbit) ----- OS2PRG CFORUM appended at 21:09:06 on 92/10/28 GMT (by HHSD012 at OS2CUST) Subject: Starting a DOS app in the foreground Ref: Append at 16:31:39 on 92/10/28 GMT (by NAAO3IY at HONE85) John, I take it you mean that a C program can start an OS/2 program in the foreground even when it is in the background. If that is the case, then you have answered my question. If that does work, I should be fixed up...Thanks Nick Gardiner HHSD012 ----- OS2PRG CFORUM appended at 23:12:43 on 92/10/28 GMT (by SC025203 at AUSVM8) Subject: Starting a DOS app in the foreground Ref: Append at 03:18:58 on 92/10/27 GMT (by HHSD012 at OS2CUST) Nick, You are making this way too complicated for what you are trying to do. I'm not sure why you need a WINAP to just start other DOS apps but I'll assume you have your reasons. This could be done as follows and with minimal (if any) coding: 1. Have a CMD file with two commands: WINAP (Your Windoze app) START WINOUT (A CMD file that WINAP writes) EXIT (Optional) 2. Have WINAP write to WINOUT.CMD, it has: START DOSAP1 (First DOS app) START DOSAP2 (Second DOS app) . . . START DOSAPn (Last DOS app) EXIT 3. Set up an icon (from templates) that starts the CMD file from number one above; or, start it in STARTUP.CMD (or any other CMD you will be running at this time). You stated that you already have the WINAP writing the DOS app names to a file, just change it to write to a CMD file (WINOUT in my example but you can name it anything) and to put "START " in front of each. The exit at the end is nice for cleanup but not absolutely necessary. Does this make sense? Stephen Poole OS/2 Service and Support - IBM Austin, TX ----- OS2PRG CFORUM appended at 02:31:53 on 92/10/29 GMT (by HHSD012 at OS2CUST) Subject: Starting a DOS app in the foreground Ref: Append at 23:12:43 on 92/10/28 GMT (by SC025203 at AUSVM8) Stephen, First, let me thank you for taking a look at this. Now on to the meat. There are plenty of reasons for a 'homegrown' windoze or OS/2 app to start a DOS app. In this particular case, we use the very powerful wordprocessing capabilities of WordPerfect and the graphics capabilities of DrawPerfect to process documents that have been put together by our Toolbook app. The data processing people try to only use DOS programs when the Windoze versions aren't quite up to snuff yet. I personally did the same thing with Personal AS/2 and WordPerfect for DOS. Now; to make it easier for the data processing people to agree to cooperate with us black sheep who are using OS/2,we had them write the command line of the DOS app to a file before they would try to execute it from Windoze. They always write to the same file handle for ease of implementation, but the command line is almost always different because of parameters passed along also. To start the Toolbook app, we start our CMD from the desktop like you suggested. It keeps running on the desktop, but unfortunately has to leave the foreground so one can interact with the Toolbook app. When the Toolbook app writes to the file handle, the CMD program 'notices' and does a START on whatever command line was written to the file. This works great except for one thing, the CMD file CANNOT throw the DOS app to the foreground unless it itself is in the foreground. You might ask why that matters, we could just switch to the window list and restore the DOS app. That is what we do now, and I am sorry to tell you that it just isn't good enough. We have some people called'users' here who will not always remember to CNTL-ESCto the Window List because when they were running Windoze they didn't have to do that. Anyway, having been told on IBMLINK that REXX wouldn't be able to start a DOS session in the foreground unless it was there, and having been told that the only way to do that was with C (or some such), that started my quest. Now I have C-Set/2 and was about to hunker down with it when the programmers at a very reputable OS/2 programming house that it couldn't be done with C either. Therefore, I am trying to get OFFICIAL word from IBM either way. Can I or Can't I force a DOS app by hook or by crook onto the foreground with an application that isn't running in the foreground. As in a previous thread, I don't care if I have to force another OS/2 program into the foreground which starts the DOS app in the foreground, because this could be done quickly enough so the user doesn't even notice. Lastly Stephen, I am sorry that I am even having to do this. If IBM had been able to throw in some sort of pseudo enhanced mode Windoze support, I probably wouldn't need to 'Switch to another session to start this program'. Then again, if Asymetrics would come out with a Toolbook 1.5 for OS/2 I wouldn'tneed pseudo enhanced mode or win-os2 for that matter since my company writes half of there apps in Toolbook. Thanks for letting me bend your eye, Nick Gardiner HHSD012 ----- OS2PRG CFORUM appended at 19:56:01 on 92/10/29 GMT (by NAAO3IY at HONE84) Subject: Starting a DOS app in the foreground Ref: Append at 21:09:06 on 92/10/28 GMT (by HHSD012 at OS2CUST) I'm shaky myself about this -- but if a C program can start CMD.EXE in the foreground, then that CMD.EXE can of course run a CMD that will start the DOS program in the foreground. John Kennedy, Hoechst Celanese (The OS/2 Hobbit)