From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Sun, 19 Jan 2003 04:48:40 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 18 ************************************************** Saturday 18 January 2003 Number 18 ************************************************** Subjects for today 1 Building PERL using Posix/2 : John Poltorak 2 Re: Building PERL using Posix/2 : John Poltorak 3 Re: UnixOS/2 Build System - mailing list : John Poltorak 4 Re: Variable argument prototypes : Stefan Neis 5 Re: Reviving posix2 mailing list : Ted Sikora 6 Re: Variable argument prototypes : Thomas E. Dickey" 7 Re: UnixOS/2 Build System - mailing list : John Poltorak 8 Reviving posix2 mailing list : John Poltorak 9 Re: Reviving posix2 mailing list : John Poltorak 10 Re: Reviving posix2 mailing list : Holger Veit 11 Re: glibc completion : Henry Sobotka 12 TERMCAP.DAT : John Poltorak 13 Syntax of bldlevel string : Thomas Hoffmann 14 Re: Syntax of bldlevel string : Yuri Dario" **= Email 1 ==========================** Date: Sun, 19 Jan 2003 11:45:43 +0000 From: John Poltorak Subject: Building PERL using Posix/2 If I wanted to attempt to build Perl using Posix/2, what would I need to do? My intention is to extract Posix/2 to \usr and keep all the EMX libs and headers under \emx and set the variables to use \usr before \emx. Is this likley to work? Do I need to doanything else? -- John **= Email 2 ==========================** Date: Sun, 19 Jan 2003 12:50:03 +0000 From: John Poltorak Subject: Re: Building PERL using Posix/2 On Sun, Jan 19, 2003 at 11:45:43AM +0000, John Poltorak wrote: > > If I wanted to attempt to build Perl using Posix/2, what would I need to > do? > > > My intention is to extract Posix/2 to \usr and keep all the EMX libs and > headers under \emx and set the variables to use \usr before \emx. > > Is this likley to work? Do I need to do anything else? ***YES***!!!! But I have no idea what... My first attempt finished quite early at this point:- `sh cflags "optimize='-O2 -fomit-frame-pointer -malign-loops=2 -malign-jumps=2 -malign-functions=2 -s'" hv.obj` -Zdll hv.c CCCMD = gcc -DPERL_CORE -c -Zomf -Zmt -DDOSISH -DOS2=2 -DEMBED -I. -D_EMX_CRT_REV_=64 -DMYTTYNAME -O2 -fomit-frame-pointer -malign-loops=2 -malign-jumps=2 -malign-functions=2 -s -Wall hv.c: In function `S_save_hek_flags': hv.c:82: warning: implicit declaration of function `offsetof' hv.c:82: parse error before `HEK' hv.c:82: parse error before `)' hv.c:79: warning: `k' might be used uninitialized in this function hv.c: In function `Perl_hv_fetch_ent': hv.c:363: parse error before `HEK' hv.c:363: parse error before `)' hv.c:362: warning: `k' might be used uninitialized in this function hv.c: In function `Perl_hv_iternext_flags': hv.c:1821: parse error before `HEK' hv.c:1821: parse error before `)' hv.c:1815: warning: `k' might be used uninitialized in this function make: *** [hv.obj] Error 1 It isn't at all obvious to me what has gone wrong at this point. One thing I have noticed is that when including the Posix/2 headers I get this additional field defined:- -DMYTTYNAME What does it signify? Is it causing a problem? -- John **= Email 3 ==========================** Date: Sun, 19 Jan 2003 12:56:29 +0000 From: John Poltorak Subject: Re: UnixOS/2 Build System - mailing list On Fri, Jan 17, 2003 at 05:47:10PM -0800, Steve Wendt wrote: > On Fri, 17 Jan 2003, John Poltorak wrote: > > > Periodically, I hope to provide summaries of builds of different apps, > > on this list but the minutiae can be very tedious for people not involved. > > I wish to express my thanks for creating the secondary list. ;) I just wish certain people would join it, otherwise I end up cluttering up this list with things most people are not particularly interested in... -- John **= Email 4 ==========================** Date: Sun, 19 Jan 2003 13:00:35 +0100 (CET) From: Stefan Neis Subject: Re: Variable argument prototypes On Sat, 18 Jan 2003, Thomas Dickey wrote: > > > programming_error (reason, arg1, arg2, arg3, arg4, arg5) > > > const char *reason; > > > { > > > char *h; > > > ... (and so on) > > > > Huh? What kind of function header syntax is that supposed to be? > > it's K&R code (pre-standard). In principle, I recognized that, I just didn't know that you can have arguments with unspecified type... Anyway, somewhere above in the original post (which I already deleted from my mailbox, so I can't check any more), there seemed to be a prototype using "..." to indicate indeterminate number of arguments. I don't think this can be mixed with the K&R style above (which requires exactly 5 more arguments after the initial one, doesn't it?). Regards, Stefan -- Micro$oft is not an answer. It is a question. The answer is 'no'. **= Email 5 ==========================** Date: Sun, 19 Jan 2003 13:23:39 -0500 From: Ted Sikora Subject: Re: Reviving posix2 mailing list John Poltorak wrote: > On Sun, Jan 19, 2003 at 05:22:51PM +0100, Holger Veit wrote: > >>On Sun, Jan 19, 2003 at 03:53:50PM +0000, John Poltorak wrote: >> >>>Are there still quite a few people subscribed to the posix2 mailing list:- ? >>> >>>http://borneo.gmd.de/mailman/listinfo/posix2 >>> >>> >>>It may be worth trying to revive it... I'm sure the work from the project >>>can be incorporated into UnixOS/2. I'd like to know if anyone on that >>>project ever tried building Perl using Posix/2 headers... >> >>I'd suggest to move this over to some kind of a developer's sublist for >>UnixOS/2 - it's after all nothing else than this. I think it makes perfect sense. UnixOS/2 is growing so dividing the work is perfect. Sure there's overlap. With OS/2 there will always be. We already have a core maintainer or base so a ux2core list would be perfect. An offshoot of posix it would deal with the base/core components of UnixOS/2. ux2bs, the build system, or ports as it really is. Then the os2-unix as the main development list. Look how cluttered os2-unix was getting. Now developers could subcribe and help out with their particular field of expertise or interest. -- Ted Sikora tsikora at ntplx.net **= Email 6 ==========================** Date: Sun, 19 Jan 2003 13:28:05 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: Variable argument prototypes On Sun, 19 Jan 2003, Stefan Neis wrote: > On Sat, 18 Jan 2003, Thomas Dickey wrote: > > > > > programming_error (reason, arg1, arg2, arg3, arg4, arg5) > > > > const char *reason; > > > > { > > > > char *h; > > > > ... (and so on) > > > > > > Huh? What kind of function header syntax is that supposed to be? > > > > it's K&R code (pre-standard). > > In principle, I recognized that, I just didn't know that you can > have arguments with unspecified type... in K&R they default to integers. > Anyway, somewhere above in the original post (which I already deleted > from my mailbox, so I can't check any more), there seemed to be a > prototype using "..." to indicate indeterminate number of arguments. > I don't think this can be mixed with the K&R style above (which requires > exactly 5 more arguments after the initial one, doesn't it?). right - the compiler should report that as an error. But that's assuming that the prototype (in a .h file) is included and made visible in the .c file where the actual function resides. Some programs do not follow this rule. There's a gcc option which is useful for finding these (-Wmissing-prototypes). That doesn't protect against having the same function declared inconsistently in different header files, of course. -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 7 ==========================** Date: Sun, 19 Jan 2003 14:34:29 +0000 From: John Poltorak Subject: Re: UnixOS/2 Build System - mailing list On Mon, Jan 20, 2003 at 01:17:38AM +1100, IanM wrote: > Hi John > > I've been away, Thought you were a bit quiet... > where is this other list, I'd hate to search > through all those posts :-( See:- http://powerusersbbs.net/mailman/listinfo/ux2bs We managed to get fairly good builds of Perl so far... Only few more test failures to iron out and I think we might be as far as we can hope to get. Ncurses also builds without a hitch as well as a few other apps. All in all, I think it provides a good base for making substantial progress. > Cheers > IanM > http://www.os2site.com/ > > > What's a 1000 lawyers at the bottom of the ocean? -- A good start -- John **= Email 8 ==========================** Date: Sun, 19 Jan 2003 15:53:50 +0000 From: John Poltorak Subject: Reviving posix2 mailing list Are there still quite a few people subscribed to the posix2 mailing list:- ? http://borneo.gmd.de/mailman/listinfo/posix2 It may be worth trying to revive it... I'm sure the work from the project can be incorporated into UnixOS/2. I'd like to know if anyone on that project ever tried building Perl using Posix/2 headers... -- John **= Email 9 ==========================** Date: Sun, 19 Jan 2003 17:12:58 +0000 From: John Poltorak Subject: Re: Reviving posix2 mailing list On Sun, Jan 19, 2003 at 05:22:51PM +0100, Holger Veit wrote: > On Sun, Jan 19, 2003 at 03:53:50PM +0000, John Poltorak wrote: > > Are there still quite a few people subscribed to the posix2 mailing list:- ? > > > > http://borneo.gmd.de/mailman/listinfo/posix2 > > > > > > It may be worth trying to revive it... I'm sure the work from the project > > can be incorporated into UnixOS/2. I'd like to know if anyone on that > > project ever tried building Perl using Posix/2 headers... > > I'd suggest to move this over to some kind of a developer's sublist for > UnixOS/2 - it's after all nothing else than this. Not sure what you're suggesting here... The Posix/2 list has been dormant for almost 18 months, but there are still probably a number of people with a good understanding of what the project can provide. I've already set up a UnixOS/2 sublist for assembling a standard build environment, to reduce noice on this list, but I'd like some input from people who are familiar with Posix/2 headers and libraries. > The focus on this > sort of 'developer' should be strictly the core package (i.e. EMX and/or > its successors), *not* dealing with the build of secondary software > (gcc, GNU software, OS/2 ports, build system, shells) to avoid duplicates > and "noise". There is a problem for a number porters of secondary software because of things like shortcomings in EMX such as the problems with strcasecmp etc which would disappear if we started using the Posix/2 headers. Since a lot of work has already gone into them, I'd like to try using them in a standard development environment. I'm assuming that whatever you do with libemu will be based on Posix/2 so I'd like to think, that when libemu came along it would slot in to wherever UnixOS/2 was at the time, but I don't want to hold up UnixOS/2 until libemu is available. > Holger > > -- > Please update your tables to my new e-mail address: > holger.veit$ais.fhg.de (replace the '$' with ' at ' -- spam-protection) > -- John **= Email 10 ==========================** Date: Sun, 19 Jan 2003 17:22:51 +0100 From: Holger Veit Subject: Re: Reviving posix2 mailing list On Sun, Jan 19, 2003 at 03:53:50PM +0000, John Poltorak wrote: > Are there still quite a few people subscribed to the posix2 mailing list:- ? > > http://borneo.gmd.de/mailman/listinfo/posix2 > > > It may be worth trying to revive it... I'm sure the work from the project > can be incorporated into UnixOS/2. I'd like to know if anyone on that > project ever tried building Perl using Posix/2 headers... I'd suggest to move this over to some kind of a developer's sublist for UnixOS/2 - it's after all nothing else than this. The focus on this sort of 'developer' should be strictly the core package (i.e. EMX and/or its successors), *not* dealing with the build of secondary software (gcc, GNU software, OS/2 ports, build system, shells) to avoid duplicates and "noise". Holger -- Please update your tables to my new e-mail address: holger.veit$ais.fhg.de (replace the '$' with ' at ' -- spam-protection) **= Email 11 ==========================** Date: Sun, 19 Jan 2003 19:55:54 -0500 From: Henry Sobotka Subject: Re: glibc completion My preference is for #2. Supplying the missing pieces seems more manageable than an entire library such as Posix/2, which already has incompatibilities with EMX (e.g. struct stat) but could be cannibalized where useful. The main disadvantage I see to #1 is the risk of EM suddenly popping up with another fix or revision of EMX, as happened the last time we got into discussion of it as abandonware. Were that to happen, we'd have two EMXes. h~ **= Email 12 ==========================** Date: Sun, 19 Jan 2003 21:15:49 +0000 From: John Poltorak Subject: TERMCAP.DAT Ncurses includes a file, terminfo.src, which, as I understand is used as a definitive source for any terminal definition you should ever need... Within it, it mentions:- #### Non-Unix Consoles # # Except for the "-emx" suffixes, these are as distributed with EMX 0.9b, # a Unix-style environment used on OS/2. (Note that the suffix makes some # names longer than 14 characters, the nominal maximum). # # Removed: rmacs=\E[10m, smacs=\E[11m, because OS/2 does not implement acs. ansi-emx|ANSI.SYS color, So presumably there is some relationship between this section and TERMCAP.DAT which come with EMX... But I don't see it at all. Can someone illuminate this relationship? I would like have a standard TERMCAP.DAT for UnixOS/2 and suggest it goes in %ETC%... This would also need a new termcap.a which has a hard coded path of /emx/etc/termcap.dat. I know this value can be overridden using the termcap variable, although I never used to know why it was necessary. BTW, I've notice another file which may be of interest to OS/2 users - emx.src. What is that used for? BTW2, Why can't we use the termcap entry for LINUX? It would be nice to have a specific definition for UnixOS/2... How doe one go about creating one? -- John **= Email 13 ==========================** Date: Sun, 19 Jan 2003 22:43:40 +0100 From: Thomas Hoffmann Subject: Syntax of bldlevel string A rather long time ago the inclusion of build information, readable with bldlevel.exe, wqas discussed. Below an "extended format" was shown and formatting restrictions were mentioned. A google search found some information and helper tools, but I did not find a clear definition of this format anywhere. Can anybody on the list tell something about this format? >NAME toolname WINDOWCOMPAT >>DESCRIPTION " at #toolname:version# at This is the UnixOS/2 tool toolname" Or the extended format: DESCRIPTION " at #Project Odin:0.5# at ##1## 23 Aug 2001 04:25:30 Univac-Job::::45 0:: at at Odin32 System DLL - Kernel32" Please note that the format of the string is pretty strict when it comes to some field lenghts. **= Email 14 ==========================** Date: Sun, 19 Jan 2003 23:05:33 +0100 (CET) From: "Yuri Dario" Subject: Re: Syntax of bldlevel string Hi, >Can anybody on the list tell something about this format? in the odin source tree, under tools\bin, you can find BldLevelInf.cmd: it can build a description string starting from option parameters. Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.quasarbbs.net/yuri * http://www.teamos2.it * http://www.opera.com/os2/ */