Date: Tue, 23 Nov 2004 00:05:22 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 455 ************************************************** Monday 22 November 2004 Number 455 ************************************************** Subjects for today 1 Re: Building Python with GCC 3.3.5 : Stefan.Neis at t-online.de 2 No controlling tty : John Poltorak 3 Re: wxWindows and gcc 3.3.5 : John Poltorak 4 Re: No controlling tty : Stefan.Neis at t-online.de 5 Re: System V Semaphore emulation : lsunley at mb.sympatico.ca 6 Re: System V Semaphore emulation : lsunley at mb.sympatico.ca 7 Re: No controlling tty : John Poltorak 8 Re: wxWindows and gcc 3.3.5 : Stefan.Neis at t-online.de 9 Re: No controlling tty : John Poltorak 10 Re: No controlling tty : Dave Yeo" 11 Symbol to refer to Innoteks GCC : Dave Yeo" 12 Re: Symbol to refer to Innoteks GCC : lsunley at mb.sympatico.ca 13 Re: No controlling tty : John Poltorak 14 Re: No controlling tty : Dave Yeo" 15 Re: No controlling tty : John Poltorak 16 Re: No controlling tty : lsunley at mb.sympatico.ca 17 Re: Symbol to refer to Innoteks GCC : Yuri Dario" 18 Re: Symbol to refer to Innoteks GCC : Steve Wendt" 19 Re: Building Python with GCC 3.3.5 : John Poltorak 20 Re: No controlling tty : John Poltorak 21 Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 : Knut St. Osmundsen" 22 Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 : Knut St. Osmundsen" 23 Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 : Knut St. Osmundsen" 24 Re: System V Semaphore emulation : Knut St. Osmundsen" 25 Re: Symbol to refer to Innoteks GCC : Knut St. Osmundsen" 26 Re: Building Python with GCC 3.3.5 : Knut St. Osmundsen" 27 Re: No controlling tty : Knut St. Osmundsen" **= Email 1 ==========================** Date: Sun, 21 Nov 2004 15:02:11 +0100 From: Stefan.Neis at t-online.de Subject: Re: Building Python with GCC 3.3.5 Hi, > gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe > -DPy_BUILD_CORE -DNDEBUG -Zomf -c config.c -o out/optimize/config.obj > In file included from ../../Include/Python.h:36, > from config.c:37: > g:/usr/include/stdlib.h:324: error: conflicting types for `intptr_t' > pyconfig.h:86: error: previous declaration of `intptr_t' Another example of an emx specific addition which is no longer needed with Innotek's libc. Removing the declaration of intptr_t from pyconfig.h should fix that, but there will be other similar problems most likely. Regards, Stefan **= Email 2 ==========================** Date: Sun, 21 Nov 2004 15:10:20 +0000 From: John Poltorak Subject: No controlling tty I've managed to build PDKSH with GCC 3.3.5 but wheb running it I always get this error msg:- sh.exe: No controlling tty (open /dev/tty: No such file or directory) Is this some sort of TERMCAP/TERMINFO problem? I'm not sure if the error msg is coming from this code in tty.c:- /* X11R5 xterm on mips doesn't set controlling tty properly - temporary hack */ # if !defined(__mips) || !(defined(_SYSTYPE_BSD43) || defined(__SYSTYPE_BSD43)) if (tfd < 0) { tty_devtty = 0; warningf(FALSE, "No controlling tty (open /dev/tty: %s)", strerror(errno)); } # endif /* __mips */ If this is the culprit would it need to be changed to work on OS/2? Or is it simply that a terminal is either not defined or sh.exe can't find that definition? -- John **= Email 3 ==========================** Date: Sun, 21 Nov 2004 15:20:58 +0000 From: John Poltorak Subject: Re: wxWindows and gcc 3.3.5 On Sat, Nov 20, 2004 at 06:57:12PM +0100, Stefan.Neis at t-online.de wrote: > Hi, > > > Is it worth trying to build the latest wxWindows with gcc 3.3.5 ? > > > > Or is it likely to end up as an exercise in frustration? > > I haven't checked yet, but since it did build with all of > gcc-2.8.1, gcc-3.2.2, and gcc-3.3.4, I would be surprised if > there are mahor problems with 3.3.5. Just wondered if you had been in contact with Dave Webster... Mail to him seems to be bouncing at the moment. > Regards, > Stefan -- John **= Email 4 ==========================** Date: Sun, 21 Nov 2004 16:22:41 +0100 From: Stefan.Neis at t-online.de Subject: Re: No controlling tty Hi, > I've managed to build PDKSH with GCC 3.3.5 but wheb running it I always > get this error msg:- > > sh.exe: No controlling tty (open /dev/tty: No such file or directory) > > Is this some sort of TERMCAP/TERMINFO problem? > > I'm not sure if the error msg is coming from this code in tty.c:- > > > /* X11R5 xterm on mips doesn't set controlling tty properly - temporary > hack */ > # if !defined(__mips) || !(defined(_SYSTYPE_BSD43) || > defined(__SYSTYPE_BSD43)) > if (tfd < 0) { > tty_devtty = 0; > warningf(FALSE, > "No controlling tty (open /dev/tty: %s)", > strerror(errno)); > } > # endif /* __mips */ > > > If this is the culprit would it need to be changed to work on OS/2? > > Or is it simply that a terminal is either not defined or sh.exe can't find > that definition? If I'm reading the code correctly, it's a warning only - so does pdksh work, or does it not work? If it does already work, modifying the above code snippet is probbly enough to silence the warning, if it does not, then Innotek's libc apparently doesn't emulate terminals/pseudo terminals the way EMX did - and there's a major problem. Regards, Stefan **= Email 5 ==========================** Date: Sun, 21 Nov 2004 09:32:18 -0500 From: lsunley at mb.sympatico.ca Subject: Re: System V Semaphore emulation In <20041121205730.D19250 at bullseye.apana.org.au>, on 11/21/04 at 09:13 PM, Andrew MacIntyre said: >On Sat, 20 Nov 2004 lsunley at mb.sympatico.ca wrote: >> I'm attempting a port of PostgreSQL version 8 (current beta 4 code) and >> one of the things that is required is a System V semaphore emulation. >> >> Does anyone on this list have one, or know of where I might find one? >I'm aware of 2: >- the daemon implementation included in Venuto's PostgreSQL v7.x port, > which is not available in any form usable for other development > (ie no headers or libs) >- a DLL based implementation I found on the web (Hobbes or LEO??), though > when I looked closely it had some serious shortcomings (AIRC - it was > quite some time ago) >I started to sketch out a daemon based implementation ala Venuto's (with >the intent of building my own PosgreSQL port) but didn't get far before >higher priorities took over... >I did ask VV whether he'd release his implementation in usable form, even >if closed source, but no positive response. I'd love to know of a viable >alternative. I'm modifying the win32 sema.c code that comes with postgreSQL v 8 to work with the OS/2 APIs for semaphore handling. The SysV semaphores seem to be placed in a list and a DosWaitMuxWaitSem can be used to wait on a list of event semaphores create with DosCreateEventSem. The trick seems to be that the Win API will call a given routine when the wait completes, which does not seem to be the case with DosWait... The Win code uses the routine to kick off some signal dispatching. Anyway, outside of the semaphore thing, there are only a few changes necessary to the makefiles and configure to obtain a build environment. There are some conflicts between typedefs in the GCC headers and those in the postgeSQL headers. I got around those by #ifdef #define _OS2EMX_H #endif at the beginning of affected routines to prevent os.h from being used. The problems are with the geometry routines and the parser backend\access\rtree\rtproc.c backend\utils\adt\geo_ops.c backend\utils\adt\geo_selfuncs.c backend\parser\keywords.c backend\parser\parser.c If anyone is interested, I'll zip up what I have for others to look at. Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 6 ==========================** Date: Sun, 21 Nov 2004 10:00:35 -0500 From: lsunley at mb.sympatico.ca Subject: Re: System V Semaphore emulation So, What was supposed to be in this message? I got a blank Lorne In <20041121121513.W84 at warpix.org>, on 11/21/04 at 12:15 PM, John Poltorak said: -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 7 ==========================** Date: Sun, 21 Nov 2004 15:59:34 +0000 From: John Poltorak Subject: Re: No controlling tty On Sun, Nov 21, 2004 at 04:22:41PM +0100, Stefan.Neis at t-online.de wrote: > If I'm reading the code correctly, it's a warning only - so does pdksh work, > or does it not work? Yes and no :-)... I get that error msgs constantly under my build environment, but copying the newly built sh.exe to a different machine, no such message appears which suggests that it is a terminal definition problem. Under my controlled environment I get this:- [G:\ux2bs\workdir\pdksh-5.2.14]ls -al README -rw-rw---a 9492 Jul 13 1999 README [G:\ux2bs\workdir\pdksh-5.2.14]sh sh: No controlling tty (open /dev/tty: No such file or directory) # ls -al README Process terminated by SIGILL - pid=0x0928 - tid=0x01 - G:\UX2BS\WORKDIR\PDKSH-5.2.14\SH.EXE [G:\ux2bs\workdir\pdksh-5.2.14]sh sh: No controlling tty (open /dev/tty: No such file or directory) # exit [G:\ux2bs\workdir\pdksh-5.2.14] TERM is set to ANSI-COLOR-2 TERMCAP is set to /etc/termcap This files does exist and contains this entry:- # nice colors for Emacs (white on blue, mode line white on cyan) ansi-color-2|ANSI.SYS color 2:\ :ti=\E[0;37;44m\E[=7l:te=\E[0m\E[=7h:\ :so=\E[1;37;46m:se=\E[0;37;44m:\ :us=\E[1;31;40m:ue=\E[0m:\ :me=\E[0;37;44m:tc=ansi: So, I'm really not sure what is missing. The new sh.exe clearly does work when some specific requirement is met but which one? > Regards, > Stefan -- John **= Email 8 ==========================** Date: Sun, 21 Nov 2004 17:51:55 +0100 From: Stefan.Neis at t-online.de Subject: Re: wxWindows and gcc 3.3.5 Hi, > Just wondered if you had been in contact with Dave Webster... Not in last two or three weeks, no ... > Mail to him seems to be bouncing at the moment. Strange ... Regards, Stefan **= Email 9 ==========================** Date: Sun, 21 Nov 2004 17:03:16 +0000 From: John Poltorak Subject: Re: No controlling tty On Sun, Nov 21, 2004 at 03:59:34PM +0000, John Poltorak wrote: > On Sun, Nov 21, 2004 at 04:22:41PM +0100, Stefan.Neis at t-online.de wrote: > > > If I'm reading the code correctly, it's a warning only - so does pdksh work, > > or does it not work? > > > Yes and no :-)... > > I get that error msgs constantly under my build environment, but copying > the newly built sh.exe to a different machine, no such message appears > which suggests that it is a terminal definition problem. After further investigtion it looks like the error msg is generated by the absence of a \dev directory on the current drive! Does that mean I need to create such a directory on every drive I wish to run sh.exe ?!!!!! Does anyone want a copy of this sh.exe to see I'm getting crazy results? -- John **= Email 10 ==========================** Date: Sun, 21 Nov 2004 11:35:47 -0800 From: "Dave Yeo" Subject: Re: No controlling tty On Sun, 21 Nov 2004 17:03:16 +0000, John Poltorak wrote: >On Sun, Nov 21, 2004 at 03:59:34PM +0000, John Poltorak wrote: >> On Sun, Nov 21, 2004 at 04:22:41PM +0100, Stefan.Neis at t-online.de wrote: >> >> > If I'm reading the code correctly, it's a warning only - so does pdksh work, >> > or does it not work? >> >> >> Yes and no :-)... >> >> I get that error msgs constantly under my build environment, but copying >> the newly built sh.exe to a different machine, no such message appears >> which suggests that it is a terminal definition problem. > > >After further investigtion it looks like the error msg is generated by the >absence of a \dev directory on the current drive! > >Does that mean I need to create such a directory on every drive I wish to >run sh.exe ?!!!!! > >Does anyone want a copy of this sh.exe to see I'm getting crazy results? I'd suggest running configure and comparing config.h against os2/config.h as there is sure to be defines that need updating or/and changing between the 2 enviroments. I tried building pdksh here and had problems with _emx_env not being defined in Innotek. I'm currently reinstalling ux2bs and will try that enviroment later. Dave **= Email 11 ==========================** Date: Sun, 21 Nov 2004 11:29:25 -0800 From: "Dave Yeo" Subject: Symbol to refer to Innoteks GCC Hi, inorder to keep software portable and compiling on both EMX and INNOTEK we need a symbol defined to differentiate the two. I'm not very creative and all I can think of is INNOTEK or __INNOTEK__ perhaps others have better ideas? eg changing #ifdef __EMX__ #define foo bar #endif to #if defined(__EMX__) && !defined(__INNOTEK__) #define foo bar #endif Once we agree on a symbol then we need Knut to automatically define the symbol in GCC Dave **= Email 12 ==========================** Date: Sun, 21 Nov 2004 13:59:18 -0500 From: lsunley at mb.sympatico.ca Subject: Re: Symbol to refer to Innoteks GCC In <20041121192951.85A4FC2F09 at generation.lgisp.net>, on 11/21/04 at 11:29 AM, "Dave Yeo" said: >Hi, inorder to keep software portable and compiling on both EMX and >INNOTEK we need a symbol defined to differentiate the two. I'm not very >creative and all I can think of is INNOTEK or __INNOTEK__ perhaps others >have better ideas? >eg changing >#ifdef __EMX__ >#define foo bar >#endif >to >#if defined(__EMX__) && !defined(__INNOTEK__) >#define foo bar >#endif >Once we agree on a symbol then we need Knut to automatically define the >symbol in GCC >Dave Is the difference covered by the version of GCC like #ifdef __GCC3__ or does it need more than that with LIBC -> EMXLIB differences? -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 13 ==========================** Date: Sun, 21 Nov 2004 20:14:29 +0000 From: John Poltorak Subject: Re: No controlling tty On Sun, Nov 21, 2004 at 11:35:47AM -0800, Dave Yeo wrote: > On Sun, 21 Nov 2004 17:03:16 +0000, John Poltorak wrote: > > >On Sun, Nov 21, 2004 at 03:59:34PM +0000, John Poltorak wrote: > >> On Sun, Nov 21, 2004 at 04:22:41PM +0100, Stefan.Neis at t-online.de wrote: > >> > >> > If I'm reading the code correctly, it's a warning only - so does pdksh work, > >> > or does it not work? > >> > >> > >> Yes and no :-)... > >> > >> I get that error msgs constantly under my build environment, but copying > >> the newly built sh.exe to a different machine, no such message appears > >> which suggests that it is a terminal definition problem. > > > > > >After further investigtion it looks like the error msg is generated by the > >absence of a \dev directory on the current drive! > > > >Does that mean I need to create such a directory on every drive I wish to > >run sh.exe ?!!!!! > > > >Does anyone want a copy of this sh.exe to see I'm getting crazy results? > > I'd suggest running configure and comparing config.h against > os2/config.h as there is sure to be defines that need updating or/and > changing between the 2 enviroments. I tried building pdksh here and had > problems with _emx_env not being defined in Innotek. I'm currently > reinstalling ux2bs and will try that enviroment later. The way PDKSH gets built is a real mess which I would like to tidy up at some point. If you are currently installing UX2BS the system may well have built PDKSH by now so you could take a look to see what I mean. As far as the requirement for Innotek is concerned, I'm developing a new UX2BS based around gcc 3.3.5 instead of 2.8.1 and have managed to get PDKSH built automatically under that, although it requires a modified diff file. I can't see anything in config.h which looks to be related to TTY. > Dave -- John **= Email 14 ==========================** Date: Sun, 21 Nov 2004 11:57:36 -0800 From: "Dave Yeo" Subject: Re: No controlling tty On Sun, 21 Nov 2004 17:03:16 +0000, John Poltorak wrote: > >After further investigtion it looks like the error msg is generated by the >absence of a \dev directory on the current drive! > >Does that mean I need to create such a directory on every drive I wish to >run sh.exe ?!!!!! EMX __open() maps /dev/tty to con. Perhaps this is missing in Innoteks gcc? Also OS/2 does have a virtual \DEV on every drive. Try dir con on a drive without /dev [N:\usr\src\pdksh-5.2.14\os2]dir con Volume in drive N has no label. The Volume Serial Number is 2962:BC14. Directory of \DEV SYS0002: The system cannot find the file specified. Same results for nul, lpt, com1 etc Dave **= Email 15 ==========================** Date: Sun, 21 Nov 2004 20:42:38 +0000 From: John Poltorak Subject: Re: No controlling tty On Sun, Nov 21, 2004 at 11:57:36AM -0800, Dave Yeo wrote: > On Sun, 21 Nov 2004 17:03:16 +0000, John Poltorak wrote: > > > > >After further investigtion it looks like the error msg is generated by the > >absence of a \dev directory on the current drive! > > > >Does that mean I need to create such a directory on every drive I wish to > >run sh.exe ?!!!!! > > EMX __open() maps /dev/tty to con. Perhaps this is missing in Innoteks > gcc? Maybe Knut could advise... > Also OS/2 does have a virtual \DEV on every drive. Try dir con on > a drive without /dev I didn't realise this... > [N:\usr\src\pdksh-5.2.14\os2]dir con > > Volume in drive N has no label. > The Volume Serial Number is 2962:BC14. > > Directory of \DEV > > > SYS0002: The system cannot find the file specified. This doesn't seem correct and doesn't agree with what you have just said. I get:- D:\>dir con Volume in drive D has no label. The Volume Serial Number is 22CA:2014. Directory of \DEV CON 0 1-01-80 12:00a 1 file(s) 0 bytes used > Same results for nul, lpt, com1 etc > Dave > -- John **= Email 16 ==========================** Date: Sun, 21 Nov 2004 15:12:17 -0500 From: lsunley at mb.sympatico.ca Subject: Re: No controlling tty In <20041121204238.H55778 at warpix.org>, on 11/21/04 at 08:42 PM, John Poltorak said: >On Sun, Nov 21, 2004 at 11:57:36AM -0800, Dave Yeo wrote: >> On Sun, 21 Nov 2004 17:03:16 +0000, John Poltorak wrote: >> >> > >> >After further investigtion it looks like the error msg is generated by the >> >absence of a \dev directory on the current drive! >> > >> >Does that mean I need to create such a directory on every drive I wish to >> >run sh.exe ?!!!!! >> >> EMX __open() maps /dev/tty to con. Perhaps this is missing in Innoteks >> gcc? >Maybe Knut could advise... >> Also OS/2 does have a virtual \DEV on every drive. Try dir con on >> a drive without /dev >I didn't realise this... >> [N:\usr\src\pdksh-5.2.14\os2]dir con >> >> Volume in drive N has no label. >> The Volume Serial Number is 2962:BC14. >> >> Directory of \DEV >> >> >> SYS0002: The system cannot find the file specified. >This doesn't seem correct and doesn't agree with what you have just >said. I get:- >D:\>dir con >Volume in drive D has no label. >The Volume Serial Number is 22CA:2014. >Directory of \DEV >CON 0 1-01-80 12:00a > 1 file(s) 0 bytes used >> Same results for nul, lpt, com1 etc >> Dave >> When I have a command window I have been building things with GCC I get [M:\psqlb4src\src\backend\utils\mb\conversion_procs]dir \nul The volume label in drive M is TheMD. The Volume Serial Number is 84CC:95C1. Directory of \DEV 1-01-80 12:00a 0 0 NUL 1 file(s) 0 bytes used With a freshly opened command window [M:\]dir nul The volume label in drive M is TheMD. The Volume Serial Number is 84CC:95C1. Directory of \DEV SYS0002: The system cannot find the file specified. Odd...... Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 17 ==========================** Date: Sun, 21 Nov 2004 23:04:26 +0100 (CET) From: "Yuri Dario" Subject: Re: Symbol to refer to Innoteks GCC On Sun, 21 Nov 2004 11:29:25 -0800, Dave Yeo wrote: >creative and all I can think of is INNOTEK or __INNOTEK__ perhaps >others have better ideas? __INNOTEK_LIBC__ (or __LIBC_INNOTEK__) is already defined Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 18 ==========================** Date: Sun, 21 Nov 2004 19:49:35 -0800 (PST) From: "Steve Wendt" Subject: Re: Symbol to refer to Innoteks GCC On Sun, 21 Nov 2004 11:29:25 -0800, Dave Yeo wrote: >Hi, inorder to keep software portable and compiling on both EMX and >INNOTEK we need a symbol defined to differentiate the two. I'm not very >creative and all I can think of is INNOTEK or __INNOTEK__ perhaps >others have better ideas? What about __OS2LIBC__ or similar? ----------- "Good people do not need laws to tell them to act responsibly, while bad people will find a way around the laws." - Plato (427-347 B.C.) **= Email 19 ==========================** Date: Mon, 22 Nov 2004 11:03:08 +0000 From: John Poltorak Subject: Re: Building Python with GCC 3.3.5 On Sun, Nov 21, 2004 at 03:02:11PM +0100, Stefan.Neis at t-online.de wrote: > Hi, > > > gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe > > -DPy_BUILD_CORE -DNDEBUG -Zomf -c config.c -o out/optimize/config.obj > > In file included from ../../Include/Python.h:36, > > from config.c:37: > > g:/usr/include/stdlib.h:324: error: conflicting types for `intptr_t' > > pyconfig.h:86: error: previous declaration of `intptr_t' > > Another example of an emx specific addition which is no longer needed with > Innotek's libc. Removing the declaration of intptr_t from pyconfig.h should > fix that, but there will be other similar problems most likely. I commented it out although there was no emx reference around it - rather something to do with win32... Once removed the compiled proceeded but stopped eventually with:- gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe -DPy_BUILD_CORE -DNDEBUG -Zomf -c ../../Objects/longobject.c -o out/optimize/longobject.obj .../../Objects/longobject.c: In function `long_format': .../../Objects/longobject.c:1141: warning: dereferencing type-punned pointer will break strict-aliasing rules .../../Objects/longobject.c: At top level: .../../Objects/longobject.c:2702: error: `long_int' redeclared as different kind of symbol g:/usr/include/sys/types.h:351: error: previous declaration of `long_int' make: *** [out/optimize/longobject.obj] Error 1 I don't see anything emx related in longobject.c. > Regards, > Stefan -- John **= Email 20 ==========================** Date: Mon, 22 Nov 2004 11:19:41 +0000 From: John Poltorak Subject: Re: No controlling tty On Sun, Nov 21, 2004 at 05:03:16PM +0000, John Poltorak wrote: > After further investigtion it looks like the error msg is generated by the > absence of a \dev directory on the current drive! For the sake expediency, I've created a \dev subdirectory just to overcome any deficiency in building PDKSH and whilst sh.exe gets built OK, make fails in the process of building the target sh.1 with this error which seems very unusual:- gcc -O -s -o sh.exe os2.o alloc.o c_ksh.o c_sh.o c_test.o c_ulimit.o edit.o emacs.o eval.o exec.o expr.o history.o io.o jobs.o lex.o mail.o main.o misc.o missing.o path.o shf.o sigact.o syn.o table.o trap.o tree.o tty.o var.o version.o vi.o -los2 ../mkman sh ./ksh.Man > tmpksh.1 Process terminated by SIGILL - pid=0x0381 - tid=0x01 - G:\UX2BS\WORKDIR\PDKSH-5.2.14\SH.EXE make: *** [sh.1] Error 127 AFAICT these are the relevant parts of the Makefile:- SHELL_PROG = sh all: $(SHELL_PROG)$(exe_suffix) $(SHELL_PROG).1 # two steps to prevent the creation of a bogus tmpksh.1 $(SHELL_PROG).1: $(srcdir)/ksh.Man $(srcdir)/mkman $(SHELL_PROG) $(srcdir)/ksh.Man > tmpksh.1 mv tmpksh.1 $(SHELL_PROG).1 Not sure what is happening here at all... -- John **= Email 21 ==========================** Date: Mon, 22 Nov 2004 13:23:50 +0100 From: "Knut St. Osmundsen" Subject: Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 Dave Yeo wrote: > On Tue, 16 Nov 2004 16:34:45 +0100, Knut St. Osmundsen wrote: > > >>Hi! >> >>There is a new GCC / LIBC alpha out: >> ftp://ftp.netlabs.org/pub/gcc/GCC-3.3.5-alpha3.zip >> >>Enjoy, >> knut >> >> >>PS. Here are the missing release notes: > > > Thanks for getting this out. Had a few things I wanted to try but wanted to do it with this > > ... > >> Known Limitiations/Problems: > > >> o Static linking not possible - will be fixed > > > I don't understand the context of this. Statically linking the LIBC into an executable or dll. Like for instance: gcc -static hello.c Will give you an unresolved external ___libc_Back_signalSomthing16bit (don't recall the exact name). Afraid I won't have time to fix this right now, but I doubt it will cause trouble for anyone since only very special applications requires staticly linked LIBC. Kind Regards, knut **= Email 22 ==========================** Date: Mon, 22 Nov 2004 13:21:24 +0100 From: "Knut St. Osmundsen" Subject: Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 Hi Dave, I will change my code to use the same EAs as the linux hpfs vfs module. It's using SYMLINK, UID, GID, RDEV and MODE from what I can see in the source code. Kind Regards, knut Dave Yeo wrote: > On Wed, 17 Nov 2004 13:36:17 -0800 (PST), Steve Wendt wrote: > > >>On Tue, 16 Nov 2004, Dave Yeo wrote: >> >> >>>Hi Knut, looking at your symlink support I notice it uses the EA LIBC.SYMLINK. >>>I wonder if it would be possible to make this work with Linux which uses the EAs >>>GID = E8 03 >>>SYMLINK = 2F 68 6F 6D 65 2F 64 61 76 65 /home/dave >>> >>>This is for a symlink from F:\home\dave\linux_home to /home/linux with my GID. >>>Linux does preserve EAs so you could have both EAs LIBC.SYMLINK and SYMLINK and set GID to 0 (which IIRC is root) or preserve GID if it is already set. >> >>Is this with HPFS, NFS, or ____? > > > HPFS though I'd imagine it would work on FAT as well. > Dave > **= Email 23 ==========================** Date: Mon, 22 Nov 2004 13:25:02 +0100 From: "Knut St. Osmundsen" Subject: Re: GCC v3.3.5 / LIBC v0.6 - Alpha 3 Dave Yeo wrote: > On Tue, 16 Nov 2004 16:34:45 +0100, Knut St. Osmundsen wrote: > > >>Hi! >> >>There is a new GCC / LIBC alpha out: >> ftp://ftp.netlabs.org/pub/gcc/GCC-3.3.5-alpha3.zip >> >>Enjoy, >> knut > > > HI Knut, trying to build the gnu file utils (old version, I don't have > the bandwidth to do much downloading) I get this error while building > dd > gcc.exe -s -Zmt -Zcrtdll -Zbin-files -Zexe -o dd dd.o ../intl/libintl.a > ../lib/libfu.a -lsocket -lgcc > dd.o: Undefined symbol _IsLower referenced from text segment > dd.o: Undefined symbol _ToUpper referenced from text segment > dd.o: Undefined symbol _IsUpper referenced from text segment > dd.o: Undefined symbol _ToLower referenced from text segment > make[1]: *** [dd] Error 1 > > Which I don't understand as you have _islower, _toupper etc defined in > libc06a3 and all the references in dd.c are lower case. Perhaps need an > alias? That's a problem with the sources you're trying to build. ToLower/ToUpper/IsUpper/ToLower have nothing to do with LIBC. All standard LIBC stuff is lowercased. Kind Regards, knut **= Email 24 ==========================** Date: Mon, 22 Nov 2004 13:34:00 +0100 From: "Knut St. Osmundsen" Subject: Re: System V Semaphore emulation lsunley at mb.sympatico.ca wrote: > Hi all, > > I'm new to this list, so please forgive me if this is a FAQ > > I'm attempting a port of PostgreSQL version 8 (current beta 4 code) and > one of the things that is required is a System V semaphore emulation. > > Does anyone on this list have one, or know of where I might find one? > > Thanks > > Lorne Sunley > Sys V sems means semctl(), semget(), semop(), right? If so, that's on the todo list for LIBC. If someone have some half decent code for this, please tell me. Same goes for the the simple shared memory interface (not the one using filehandles and mmap). Kind Regards, knut **= Email 25 ==========================** Date: Mon, 22 Nov 2004 13:40:25 +0100 From: "Knut St. Osmundsen" Subject: Re: Symbol to refer to Innoteks GCC From ReleaseNotes.os2: v3.2.2 Beta 4: -------------- [snip] o GCC defines __INNOTEK_LIBC__=0x005. The value reflects the LIBC version, form: 0xGMM Where G=major version and MM=middle version numbers. The value for LIBC v0.6 is 0x006. Kind Regards, knut Dave Yeo wrote: > Hi, inorder to keep software portable and compiling on both EMX and > INNOTEK we need a symbol defined to differentiate the two. I'm not very > creative and all I can think of is INNOTEK or __INNOTEK__ perhaps > others have better ideas? > eg changing > #ifdef __EMX__ > #define foo bar > #endif > to > #if defined(__EMX__) && !defined(__INNOTEK__) > #define foo bar > #endif > > Once we agree on a symbol then we need Knut to automatically define the > symbol in GCC > Dave > > > > **= Email 26 ==========================** Date: Mon, 22 Nov 2004 13:44:30 +0100 From: "Knut St. Osmundsen" Subject: Re: Building Python with GCC 3.3.5 John Poltorak wrote: > On Sun, Nov 21, 2004 at 03:02:11PM +0100, Stefan.Neis at t-online.de wrote: > >> Hi, >> >> >>>gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe >>>-DPy_BUILD_CORE -DNDEBUG -Zomf -c config.c -o out/optimize/config.obj >>>In file included from ../../Include/Python.h:36, >>> from config.c:37: >>>g:/usr/include/stdlib.h:324: error: conflicting types for `intptr_t' >>>pyconfig.h:86: error: previous declaration of `intptr_t' >> >>Another example of an emx specific addition which is no longer needed with >>Innotek's libc. Removing the declaration of intptr_t from pyconfig.h should >>fix that, but there will be other similar problems most likely. > > > I commented it out although there was no emx reference around it - rather > something to do with win32... > > Once removed the compiled proceeded but stopped eventually with:- > > > gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe > -DPy_BUILD_CORE -DNDEBUG -Zomf -c ../../Objects/longobject.c -o > out/optimize/longobject.obj > ../../Objects/longobject.c: In function `long_format': > ../../Objects/longobject.c:1141: warning: dereferencing type-punned > pointer will break strict-aliasing rules > ../../Objects/longobject.c: At top level: > ../../Objects/longobject.c:2702: error: `long_int' redeclared as different > kind of symbol > g:/usr/include/sys/types.h:351: error: previous declaration of `long_int' > make: *** [out/optimize/longobject.obj] Error 1 > > > I don't see anything emx related in longobject.c. Old problem, blame IBM. Heres the sys/types.h code: /* bird: OS/2 Toolkit Pollution. */ #if !defined (_POSIX_SOURCE) || defined(__USE_EMX) #ifdef __32BIT__ typedef int long_int; /* 32-bit compilers */ #else typedef long int long_int; /* 16-bit compilers */ #endif #endif /* not _POSIX_SOURCE */ I'll change it to long int on both platforms and perhaps try hide it by some additional __USE_IBM_UGLYNESS preprocessor check or something. Kind Regards, knut **= Email 27 ==========================** Date: Mon, 22 Nov 2004 13:54:36 +0100 From: "Knut St. Osmundsen" Subject: Re: No controlling tty Stefan.Neis at t-online.de wrote: > Hi, > > >>I've managed to build PDKSH with GCC 3.3.5 but wheb running it I always >>get this error msg:- >> >>sh.exe: No controlling tty (open /dev/tty: No such file or directory) >> >>Is this some sort of TERMCAP/TERMINFO problem? >> >>I'm not sure if the error msg is coming from this code in tty.c:- >> >> >>/* X11R5 xterm on mips doesn't set controlling tty properly - temporary >>hack */ >># if !defined(__mips) || !(defined(_SYSTYPE_BSD43) || >>defined(__SYSTYPE_BSD43)) >> if (tfd < 0) { >> tty_devtty = 0; >> warningf(FALSE, >> "No controlling tty (open /dev/tty: %s)", >> strerror(errno)); >> } >># endif /* __mips */ >> >> >>If this is the culprit would it need to be changed to work on OS/2? >> >>Or is it simply that a terminal is either not defined or sh.exe can't find >>that definition? > > > If I'm reading the code correctly, it's a warning only - so does pdksh work, > or does it not work? If it does already work, modifying the above code > snippet is probbly enough to silence the warning, if it does not, then Innotek's > libc apparently doesn't emulate terminals/pseudo terminals the way EMX did - and > there's a major problem. IIRC EMX implements /dev/tty. LIBC doesn't do that yet. I'm not sure what the requirements for getting that working 100% are, and it's gonna take some time investigating it. I ran into the concept of controlling tty while looking at process groups / job control / signals but haven't had time to delve into all the details yet. Any help on the subject would be appreciated, especially if it includes code. :-) Kind Regards, knut