Date: Fri, 7 May 2004 00:04:22 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 371 ************************************************** Thursday 06 May 2004 Number 371 ************************************************** Subjects for today 1 Re: Emacs patches for v20.6 : John Poltorak 2 Re: OpenOffice : Akira Hatakeyama 3 Re: OpenOffice : John Poltorak 4 Re: timestamps in tar files : Thomas Dickey 5 Re: timestamps in tar files : Thomas Dickey 6 Re: timestamps in tar files : John Poltorak 7 Re: OpenOffice : Sergey Yevtushenko 8 Re: OpenOffice : John Poltorak 9 Re: OpenOffice : Sergey Yevtushenko 10 Re: timestamps in tar files : Thomas Dickey 11 Re: OpenOffice : Stefan.Neis at t-online.de 12 Re: OpenOffice : Neil Waldhauer" 13 OOO_1_1_1 : John Poltorak 14 Re: OpenOffice : Stefan.Neis at t-online.de 15 vxcron : John Poltorak 16 Re: OpenOffice : Sergey Yevtushenko 17 Re: OpenOffice : Andreas Buening 18 (no subject) : Jerry Garren 19 Re: OpenOffice : Dave Saville" 20 Re: OpenOffice : John Poltorak 21 distcc : John Poltorak 22 Re: OOO_1_1_1 : Sebastian Wittmeier" 23 Serious man problem : Alex Newman" 24 Re: OOO_1_1_1 : Alex Newman" 25 Re: Re: Building EMACS v20.7 : Arnstein Prytz 26 Re: distcc : Steve Wendt 27 man problem fixed : Alex Newman" 28 Re: Building EMACS v20.7 : Masaru Nomiya 29 Re: Building EMACS v20.7 : Masaru Nomiya 30 Re: OpenOffice : Stefan.Neis at t-online.de 31 Re: distcc : Stefan.Neis at t-online.de 32 Re: OOO_1_1_1 : John Poltorak 33 Re: distcc : John Poltorak 34 Re: OOO_1_1_1 : John Poltorak 35 Re: distcc : Stefan.Neis at t-online.de 36 Re: distcc : Sergey Yevtushenko 37 Re: distcc : John Poltorak 38 Alternative to 'tar zxf' : John Poltorak 39 Re: Alternative to 'tar zxf' : Thomas Dickey 40 Re: Alternative to 'tar zxf' : Illya Vaes 41 Re: distcc : Stefan.Neis at t-online.de 42 Re: distcc : Sergey Yevtushenko 43 Re: Alternative to 'tar zxf' : T.Sikora" **= Email 1 ==========================** Date: Wed, 5 May 2004 15:28:41 +0100 From: John Poltorak Subject: Re: Emacs patches for v20.6 On Wed, May 05, 2004 at 01:22:15PM +0200, Christian Hennecke wrote: > On Wed, 5 May 2004 11:41:06 +0100, John Poltorak wrote: > > >Does anyone know where I can find the OS/2 patches of emacs 20.6? > > > >I would like to compare them against those in 20.7 and if possible > >consolidate them. > > I don't think that Jeremy provided patches, but the complete source > code. It's available at Hobbes: > http://hobbes.nmsu.edu/pub/os2/apps/editors/emacs/e206src.zip Is that the only archive required to build emacs? > Christian Hennecke -- John **= Email 2 ==========================** Date: Thu, 6 May 2004 00:25:18 +0900 From: Akira Hatakeyama Subject: Re: OpenOffice Greetings. On Tue, May 04, 2004 at 04:24:47PM +0100, John Poltorak wrote: > > I very much doubt that you would find enough people to fund an > > OpenOffice port to XFree86/OS2 > > Who is to say it actually needs porting? > AFAIK, problem is build system, not code itself. OOo build system requires up to date GCC, and STL port. Now Innotech released up to date version of the GCC, but STL port is not ready yet (VAC version exist, but no GCC version). > If it just uses X11 it might build straight out of the box. Unlikely, I > know, but building Unix apps on OS/2 has become much easier in recent > years. X11 version would be first step for the OS/2 native version. But still build system/tools are major problem of OS/2 port. -- Akira Hatakeyama E-Mail: akira at sra.co.jp http://www.sra.co.jp/people/akira/index.html chigasaki-minami, tsuzuki ward, yokohama, japan **= Email 3 ==========================** Date: Wed, 5 May 2004 16:25:36 +0100 From: John Poltorak Subject: Re: OpenOffice On Thu, May 06, 2004 at 12:25:18AM +0900, Akira Hatakeyama wrote: > Greetings. > > On Tue, May 04, 2004 at 04:24:47PM +0100, John Poltorak wrote: > > > I very much doubt that you would find enough people to fund an > > > OpenOffice port to XFree86/OS2 > > > > Who is to say it actually needs porting? > > > > AFAIK, problem is build system, not code itself. > OOo build system requires up to date GCC, and STL port. > Now Innotech released up to date version of the GCC, > but STL port is not ready yet (VAC version exist, but > no GCC version). Where do I find the old STL port? > -- > Akira Hatakeyama E-Mail: akira at sra.co.jp > http://www.sra.co.jp/people/akira/index.html > chigasaki-minami, tsuzuki ward, yokohama, japan -- John **= Email 4 ==========================** Date: Wed, 5 May 2004 11:29:52 -0400 (EDT) From: Thomas Dickey Subject: Re: timestamps in tar files On Wed, 5 May 2004, John Poltorak wrote: > > > How do I get a listing of files including sizes and timestamps in a tar > archive? > > Does tar adjust timestamps according to TZ? In doing a comparison of > zipped and gzipped archives the times on the same files are often out by > exactly one hour, and I can't tell if that difference is accounted for by > the tar extract. Does anyone know? when I notice this, it's usually a problem with zip rather than tar. -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net **= Email 5 ==========================** Date: Wed, 5 May 2004 11:29:52 -0400 (EDT) From: Thomas Dickey Subject: Re: timestamps in tar files On Wed, 5 May 2004, John Poltorak wrote: > > > How do I get a listing of files including sizes and timestamps in a tar > archive? > > Does tar adjust timestamps according to TZ? In doing a comparison of > zipped and gzipped archives the times on the same files are often out by > exactly one hour, and I can't tell if that difference is accounted for by > the tar extract. Does anyone know? when I notice this, it's usually a problem with zip rather than tar. -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net **= Email 6 ==========================** Date: Wed, 5 May 2004 16:36:52 +0100 From: John Poltorak Subject: Re: timestamps in tar files On Wed, May 05, 2004 at 11:29:52AM -0400, Thomas Dickey wrote: > On Wed, 5 May 2004, John Poltorak wrote: > > > > > > > How do I get a listing of files including sizes and timestamps in a tar > > archive? > > > > Does tar adjust timestamps according to TZ? In doing a comparison of > > zipped and gzipped archives the times on the same files are often out by > > exactly one hour, and I can't tell if that difference is accounted for by > > the tar extract. Does anyone know? > > when I notice this, it's usually a problem with zip rather than tar. You can see timestamps within zipfiles by listing the files (-l). Those timestamps change according to TZ sometimes - depending on which version of unzip you have. I don't know of an equivalent way of checking timestamps in a tar archive. Is there such an option? > -- > Thomas E. Dickey > http://invisible-island.net > ftp://invisible-island.net -- John **= Email 7 ==========================** Date: Wed, 05 May 2004 18:46:23 +0300 From: Sergey Yevtushenko Subject: Re: OpenOffice Akira, >>Who is to say it actually needs porting? >> > > AFAIK, problem is build system, not code itself. > OOo build system requires up to date GCC, and STL port. > Now Innotech released up to date version of the GCC, > but STL port is not ready yet (VAC version exist, but > no GCC version). Recent versions of GCC have STL included and I doubt that OO requires more that just STL. But even if this is true, GCC version of STLport also exist. I think that GCC is one of the "main" compilers for the STLport developers. So most likely there is no problems with OO build tools under OS/2. Regards, Sergey. -- *--------------------------------------------- ES at Home http://es.os2.ru/ **= Email 8 ==========================** Date: Wed, 5 May 2004 16:57:06 +0100 From: John Poltorak Subject: Re: OpenOffice On Wed, May 05, 2004 at 06:46:23PM +0300, Sergey Yevtushenko wrote: > Recent versions of GCC have STL included and I doubt that OO requires > more that just STL. But even if this is true, GCC version of STLport > also exist. I think that GCC is one of the "main" compilers for the > STLport developers. So most likely there is no problems with OO build > tools under OS/2. Has anyone actually tried to build it on OS/2? It would be useful to catalogue any errors encountered. > Regards, > Sergey. > -- > *--------------------------------------------- > ES at Home http://es.os2.ru/ -- John **= Email 9 ==========================** Date: Wed, 05 May 2004 19:29:17 +0300 From: Sergey Yevtushenko Subject: Re: OpenOffice John, > > Has anyone actually tried to build it on OS/2? If you mean OO then I don't know. I've just looked into sources, but not tried to build it. Regards, Sergey. -- *--------------------------------------------- ES at Home http://es.os2.ru/ **= Email 10 ==========================** Date: Wed, 5 May 2004 12:22:53 -0400 (EDT) From: Thomas Dickey Subject: Re: timestamps in tar files On Wed, 5 May 2004, John Poltorak wrote: > On Wed, May 05, 2004 at 11:29:52AM -0400, Thomas Dickey wrote: > > On Wed, 5 May 2004, John Poltorak wrote: > > > > > > > > > > > How do I get a listing of files including sizes and timestamps in a tar > > > archive? > > > > > > Does tar adjust timestamps according to TZ? In doing a comparison of > > > zipped and gzipped archives the times on the same files are often out by > > > exactly one hour, and I can't tell if that difference is accounted for by > > > the tar extract. Does anyone know? > > > > when I notice this, it's usually a problem with zip rather than tar. > > You can see timestamps within zipfiles by listing the files (-l). Those > timestamps change according to TZ sometimes - depending on which version > of unzip you have. > > I don't know of an equivalent way of checking timestamps in a tar > archive. Is there such an option? tar tvf foo (the "v" turns on the verbose listing of timestamps, etc). -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net **= Email 11 ==========================** Date: Wed, 05 May 2004 18:47:51 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: OpenOffice Akira Hatakeyama schrieb: > Now Innotech released up to date version of the GCC, > but STL port is not ready yet (VAC version exist, but > no GCC version). Sorry? Normally, STL is part of GCC distribution, no need to port anything... If gcc works, STL does automatically work as well. Regards, Stefan **= Email 12 ==========================** Date: Wed, 5 May 2004 09:45:55 -0700 From: "Neil Waldhauer" Subject: Re: OpenOffice on Thu, 6 May 2004 00:25:18 +0900, Akira Hatakeyama wrote: > Now Innotech released up to date version of the GCC, > but STL port is not ready yet (VAC version exist, but > no GCC version). I tested the Standard Library features of GCC for compliance with ISO ANSI C++ by compiling the examples from Advanced C++. The Innotek port complies with the standard. I guess if you need support for the now-deprecated STL, then STLPort is the way to go. Neil Neil Waldhauer neil at blondeguy.com www.blondeguy.com Expert consulting for OS/2 and eComStation To steal ideas from one person is plagiarism; to steal from many is research. **= Email 13 ==========================** Date: Wed, 5 May 2004 18:57:51 +0100 From: John Poltorak Subject: OOO_1_1_1 If anyone is interested in trying to build Open Office can we agree an a release version? I suggest OOO_1_1_1:- http://ooo.ximian.com/packages/OOO_1_1_1/OOO_1_1_1.tar.bz2 That dates from 22 Mar 2004 and is 146MB in size. In the same directory there are a number of other files called ooo-build-*. Does anyone know what they are? -- John **= Email 14 ==========================** Date: Wed, 05 May 2004 18:51:13 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: OpenOffice Sergey Yevtushenko schrieb: > Recent versions of GCC have STL included and I doubt that > OO requires > more that just STL. But even if this is true, GCC version > of STLport > also exist. Template instantiation might be a problem with STL/gcc on non-ELF platforms? But I suppose the worst that will happen is that the executable just gets unreasonably big/ compilation time gets unreasonably long, iff the same stuff is instantiated over and over again. Regards, Stefan **= Email 15 ==========================** Date: Wed, 5 May 2004 19:32:39 +0100 From: John Poltorak Subject: vxcron ISTR reading about an OS/2 port of vxcron a while back. Does anyone know if the source is available anywhere? -- John **= Email 16 ==========================** Date: Wed, 05 May 2004 21:31:27 +0300 From: Sergey Yevtushenko Subject: Re: OpenOffice Stefan, > >>Recent versions of GCC have STL included and I doubt that >>OO requires >>more that just STL. But even if this is true, GCC version >>of STLport >>also exist. > > > Template instantiation might be a problem with > STL/gcc on non-ELF platforms? But I suppose > the worst that will happen is that the > executable just gets unreasonably big/ > compilation time gets unreasonably long, iff > the same stuff is instantiated over and over > again. As far as I know, there is a workaround for this in current port of GCC, but I didn't tested how effective it is. At least compilation of applications such as AStyle gives very reasonable executable size (about 160K after LXLite). Regards, Sergey. -- *--------------------------------------------- ES at Home http://es.os2.ru/ **= Email 17 ==========================** Date: Wed, 05 May 2004 20:49:23 +0200 From: Andreas Buening Subject: Re: OpenOffice Stefan.Neis at t-online.de wrote: > Template instantiation might be a problem with > STL/gcc on non-ELF platforms? But I suppose > the worst that will happen is that the > executable just gets unreasonably big/ > compilation time gets unreasonably long, iff > the same stuff is instantiated over and over > again. -fno-implicit-templates or -fno-implicit-inline-templates will tell you. ;-) Bye, Andreas **= Email 18 ==========================** Date: Wed, 05 May 2004 12:03:40 -0700 From: Jerry Garren Subject: (no subject) unsubscribe **= Email 19 ==========================** Date: Wed, 05 May 2004 21:15:49 +0100 (BST) From: "Dave Saville" Subject: Re: OpenOffice Just a thought John I wonder if you might need to lie about what OS? I mean maybe there are still some OS/2 if defs or similar for PM buried in the bowels of the code. We could most likely live with a driveless build - Like perl it ought to run from any drive that has the dir tree on it. -- Regards Dave Saville **= Email 20 ==========================** Date: Wed, 5 May 2004 21:59:29 +0100 From: John Poltorak Subject: Re: OpenOffice On Wed, May 05, 2004 at 09:15:49PM +0100, Dave Saville wrote: > Just a thought John > > I wonder if you might need to lie about what OS? I mean maybe there > are still some OS/2 if defs or similar for PM buried in the bowels of > the code. We could most likely live with a driveless build - Like > perl it ought to run from any drive that has the dir tree on it. I have no idea about any residual OS/2 support which may exist in OO. I've just downloaded the source and am still trying to extract it. That is taking longer than the download of the 150MB archive! The first priority must be to get something, anything, built and then take it from there. There must be a lot of options to try before finding the optimum build method. A driveless build would be OK for me if that's all I can have. > -- > Regards > > Dave Saville > -- John **= Email 21 ==========================** Date: Wed, 5 May 2004 22:12:20 +0100 From: John Poltorak Subject: distcc I've been trying to build distcc, see:- http://distcc.samba.org/ with the aim of providing a distributed compiling environment for a large project such as building Open Office. It partly builds straight out of the box - at least I get a runnable distcc.exe, but it also requires a daemon - distccd, which doesn't build - it fails with:- c:\ux2bs\posix2\lib/cExt.a(itimer.o): Undefined symbol __beginthread referenced from text segment make: *** [distccd.exe] Error 1 Hopefully this can be resolved by using appropriate *FLAGS. Just wondered if anyone had any experience of using distcc to spread a build over multiple computers... I can't really see how it should be configured. -- John **= Email 22 ==========================** Date: Thu, 06 May 2004 01:02:57 +0200 (CEST) From: "Sebastian Wittmeier" Subject: Re: OOO_1_1_1 On Wed, 5 May 2004 18:57:51 +0100, John Poltorak wrote: >If anyone is interested in trying to build Open Office can we agree an a >release version? > >I suggest OOO_1_1_1:- > >http://ooo.ximian.com/packages/OOO_1_1_1/OOO_1_1_1.tar.bz2 cool, parts of the source still have German comments :-) Sebastian **= Email 23 ==========================** Date: Thu, 06 May 2004 09:23:23 +1000 (EST) From: "Alex Newman" Subject: Serious man problem Hi there, I used to have a working man setup, but after installing groff 1.17 and some other stuff, man is dead as a doornail. This is playing merry hell with my xman setup too (of course) - which I badly need. I've trawled Leo, hobbes, os2ports, unixos2, etc., but haven't found anything recent. The output to man is: ............................................. [P:\Hold]man ls groff: can't find `DESC' file groff:fatal error: invalid device `nippon' (END) Process terminated by SIGPIPE Process terminated by SIGPIPE ............................................. The output to man -d is: ............................................. [P:\Hold]man -d ls Reading config file p:/etc/man.conf using p:\usr\bin\less.exe as pager adding p:/usr/man to manpath adding p:/usr/x11r6/man to manpath adding p:/XFree86/man to manpath adding E:/lib/R/man to manpath will try to write p:/usr/man/cat1/ls.1.bz2 if needed status from is_newer() = -2 using default preprocessor sequence found tbl(1) directive not executing command: cat p:/usr/man/man1/ls.1 | tbl | nroff -mandoc -Tnippon | p:\usr\bin\less.exe ............................................. Something is forcing nroff/groff use -Tnippon: I've tried commenting out the -Tnippon in the various associated cmd/sh scripts, but can't shake it: it seems to be hard coded. I don't have the knowledge to sort this thing out (although I have spent a good deal of time trying to find out). As far as I can tell, the DESC files are all in their right places - I haven't moved them since I unpacked them. Maybe my directory structure is stuffed up (e.g., put groff too far down the tree); or I have mismatched versions of various things (text/file utils, groff suite, man version, etc., etc.); or some weird setting in config.sys what have I forgotten about. Moving groff support files up a level (into /usr/share/groff, and resetting the env variables), however, does nothing to fix this Config.sys settings are: REM *** Start groff and manpath environment lines *** SET MANPATH=p:/usr/man;p:/usr/x11r6/man;p:/XFree86/man;E:/XEmacs/man;E:/Perl/MAN;E:/lib/R/man; SET MAN_CONF=p:/etc/man.conf SET GROFF_TMAC_PATH=p:/usr/share/groff/1.17/tmac SET GROFF_FONT_PATH=p:/usr/share/groff/1.17/font SET GROFF_TMPDIR=h:/tmp/tmp REM SET GROFF_HYPHEN=p:/usr/share/groff/1.17/tmac REM SET GROFF_TYPESETTER=ascii REM SET REFER=p:/usr/share/groff/1.17/dict/papers/index REM *** End groff and manpath environment lines *** Note the explicit "groff_typesetter" variable (ascii). There is probably a trivially simple solution to this, but it's eluding me. I need some help! If anyone can shed some light on this, please email me (on or off list, whichever preferred). I've got some nasty bruises where I keep banging my head on the desk. Regards, Alex Newman **= Email 24 ==========================** Date: Thu, 06 May 2004 09:33:23 +1000 (EST) From: "Alex Newman" Subject: Re: OOO_1_1_1 On Wed, 5 May 2004 18:57:51 +0100, John Poltorak wrote: > > If anyone is interested in trying to build Open Office can we agree an a > release version? I would try to help, but my attempts to build even simple things (ask Dave Yeo ;) have mostly met with failure, so attempting to contribute to 000111 would be out of the question. Having said that, with my experiences of the (registered) Innotek package of 000111 a native OS/2 version is desparately needed (I can't use 000111 in a production situation - far too dangerous). I could put my hand up to do some testing of compiled stuff in about 8 weeks time, though. Anything else would require a complete overhaul of my unisox2 environment, for which I haven't the time (it was the switch from my old emx tree over to a pure unixos2 one that coincided with the nobbling of my 'man' setup). Regards, Alex. **= Email 25 ==========================** Date: Thu, 6 May 2004 09:58:17 +1000 From: Arnstein Prytz Subject: Re: Re: Building EMACS v20.7 Dear John, > > John> I think the %NAME% variable is only required at runtime. > > No. It's required when compiling Emacs 20.7 for OS/2. > Well I made a specific point of not setting it and it still built without > any errors. I've tried it again just to confirm that this is the case. Without setting NAME I was unable to complete the build process. I do have HOME, USER, USERFULLNAME and LOGNAME set as well, and that was not enough. > > Arnstein> make[1]: Entering directory `E:/var/tmp/emacs/emacs-20.7/leim' > > Arnstein> ../src/emacs.exe -batch --no-init-file --no-site-file -l ../lisp/international/titdic-cnv \ I can understand why NAME is needed for the build as well, because, in order to compile the leim stuff, emacs.exe is called, which presumably needs NAME irrespective of the input flags. I don't know why yours works. Have you tried to compile by adding set NAME= to the start of the build script you posted? That would convince me that somehow your system setup needs different variables set than Masaru's and me. Regards, Arnstein ---------------------------------------------------------------------------- Arnstein Prytz Arnstein.Prytz at jcu.edu.au School of Maths and Physics ph: 61-7-47815183 James Cook University fax: 61-7-47815880 Townsville, Queensland 4811, Australia ---------------------------------------------------------------------------- **= Email 26 ==========================** Date: Wed, 5 May 2004 17:50:29 -0700 (PDT) From: Steve Wendt Subject: Re: distcc On Wed, 5 May 2004, John Poltorak wrote: > c:\ux2bs\posix2\lib/cExt.a(itimer.o): Undefined symbol __beginthread > referenced from text segment Perhaps you need to link with pthreads. **= Email 27 ==========================** Date: Thu, 06 May 2004 11:14:32 +1000 (EST) From: "Alex Newman" Subject: man problem fixed ...rtfm... And it was a simple fix. I added: NEQN neqn -Tascii NROFF nroff -mandoc -Tascii TROFF groff -Tps -mandoc to man.conf after reading Jun's readme. All working perfectly again... Apologies for the wasted bandwidth. Regards, Alex. **= Email 28 ==========================** Date: Thu, 06 May 2004 10:52:10 +0900 From: Masaru Nomiya Subject: Re: Building EMACS v20.7 Hello, In the Message; Subject : Re: Building EMACS v20.7 Message-ID : <20040505110053.H85 at warpix.org> Date & Time: Wed, 5 May 2004 11:00:53 +0100 [John] == John Poltorak has written: John> But first I'd like to see if it is possible to get a patch which includes John> any patches that were made for v19.33 and v20.6. It appears that a number John> of people revert back to 19.33 because of errors in newer versions. There exists a pacth (em202os2.tar.gz) for Emacs 20.2 at http://www.bekkoame.ne.jp/~s-osamu/ We don't know where he is, but HP remains,,,, John> In 20.7, I am unable to read INFO files - emacs simply exits. Wrote in your .emacs as follows; ;; Info files (setq Info-default-directory-list (append Info-default-directory-list (list (expand-file-name "g:/emx/info")))) # I place all info files at g:/emx/info # You must edit 'dir' file in your info files's directory. Then input M-x info on the Emacs. 1.How to make info file? Read *.texi file into Emacs's buffa by C-x C-f, then M-x texinfo-format-buffer[RET} Me> PS. I'm using Yatex, emacs-w3m (this requires w3m), etc.... John> What is w3m and Yatex? Wanderlust is very nice! This is not only a MUA, but also Newsgroup reader. Moreover we can read news accounts which are on the news agency HP (ex. BBC, but mainly Japanese news agency's). For Wanderlust, see http://www.gohome.org/wl/index.html emacs-w3m http://emacs-w3m.namazu.org/index-en.html w3m http://www2u.biglobe.ne.jp/~hsaka/w3m/index.html yatex http://www.yatex.org/ This maybe useful for you. > Arnstein lookup http://lookup.sourceforge.net/ Regards, --- Masaru Nomiya mail-to: nomiya at ttmy.ne.jp "No Windows, no gains!" ..... "Why, I am wrong?" -- Bill -- **= Email 29 ==========================** Date: Thu, 06 May 2004 11:40:15 +0900 From: Masaru Nomiya Subject: Re: Building EMACS v20.7 Hello, In the Message; Subject : Re: Building EMACS v20.7 Message-ID : <200405042304.JAA022.25 at goanna.jcu.edu.au> Date & Time: Wed, 5 May 2004 09:04:19 +1000 [Arnstein] == Arnstein Prytz has written: Arnstein> (setq default-frame-alist [...] Arnstein> ; Spinifex: AcerView 79g, 1200x1024, 17" monitor, 11 pt [...] Arnstein> (font . "11.Courier") Arnstein> (menu-font . "11.Times New Roman")))) Arnstein> ; Goanna: Benq FP991, 1200x1024, 19" LCD monitor, 9 pt [...] Arnstein> (font . "10.Courier") Arnstein> (menu-font . "9.Times New Roman")))) Arnstein> ; Gecko: IBM Thinkpad T41 laptop, 1400x1050, 12 pt [...] Arnstein> (font . "11.Courier") Arnstein> (menu-font . "11.Times New Roman")))) I can't understand your font settings. That is. we can use 14pt, 16pt, or 24pt fonts in Emacs. If we want to use aother size of font, we must generate it and describe it in .Xdefaults, I think. So, Mr. Sasaki provide us the tool, outline.exe, to generate fot this purpose. The outline.exe is at ../emacs-20.7/os2/fonts. Regards, --- Masaru Nomiya mail-to: nomiya at ttmy.ne.jp "Bill! You married with Computers. Not with Me!" "No..., with money." **= Email 30 ==========================** Date: Thu, 06 May 2004 08:57:25 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: OpenOffice Andreas Buening schrieb: > Stefan.Neis at t-online.de wrote: > > -fno-implicit-templates or > -fno-implicit-inline-templates > will tell you. ;-) That's what I'm using for my own code, but it requires adding code for explicit instantiation, so that's not something I'd like to add for a big application that's heavily using STL. Regards, Stefan **= Email 31 ==========================** Date: Thu, 06 May 2004 09:17:08 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: distcc John Poltorak schrieb: > which doesn't build - > it fails with:- > > c:\ux2bs\posix2\lib/cExt.a(itimer.o): Undefined symbol > __beginthread > referenced from text segment > make: *** [distccd.exe] Error 1 Trying to compile a multithreaded program without -Zmt? Try adding -Zmt -D__ST_MT_ERRNO__ to both CFLAGS (or even CPPFLAGS, so C++ gets it as well) and LDFLAGS. Oh, and start with a "make clean" after that change, otherwise linking might even succeed but the individual object files would still be compiled with single-threaded options. Regards, Stefan **= Email 32 ==========================** Date: Thu, 6 May 2004 09:46:07 +0100 From: John Poltorak Subject: Re: OOO_1_1_1 On Thu, May 06, 2004 at 09:33:23AM +1000, Alex Newman wrote: > On Wed, 5 May 2004 18:57:51 +0100, John Poltorak wrote: > > > > > If anyone is interested in trying to build Open Office can we agree an a > > release version? > > I would try to help, but my attempts to build even simple things You should try using UX2BS and see if you can build Perl. > I could put my hand up to do some testing of compiled stuff in about 8 weeks time, though. Maybe we can manage to set up a distributed build environment in that time so having remote systems for doing some compiling might be useful. > Anything else would > require a complete overhaul of my unisox2 environment, for which I haven't the time (it was the switch from my > old emx tree over to a pure unixos2 one that coincided with the nobbling of my 'man' setup). There are usually problems in trying to put things in a Unix-like environment is they were not originally built that way. That is one of the reasons I'm trying to build everything from scratch - so that it all works together harmoniously. Well, that's the theory anyway. > Regards, > Alex. > -- John **= Email 33 ==========================** Date: Thu, 6 May 2004 10:06:20 +0100 From: John Poltorak Subject: Re: distcc On Wed, May 05, 2004 at 05:50:29PM -0700, Steve Wendt wrote: > On Wed, 5 May 2004, John Poltorak wrote: > > > c:\ux2bs\posix2\lib/cExt.a(itimer.o): Undefined symbol __beginthread > > referenced from text segment > > Perhaps you need to link with pthreads. I may need to at some point but this particular problem was resolved by setting:- CFLAGS=-Zomf -O2 -s -Zmt -D__ST_MT_ERRNO__ So now I have a distccd which appears to build and has produced a runnable exe, but when I run it, it always returns with:- distccd[12738] (main) Warning: nice 5 failed: Invalid argument distccd[12738] (dcc_discard_root) CRITICAL! still have root privileges after trying to discard them! distccd[12738] (dcc_exit) exit: code 112; self: 0.000000 user 0.000000 sys; children: 0.000000 user 0.000000 sys Looks like some dabbling with code may be required... -- John **= Email 34 ==========================** Date: Thu, 6 May 2004 10:21:40 +0100 From: John Poltorak Subject: Re: OOO_1_1_1 On Thu, May 06, 2004 at 01:02:57AM +0200, Sebastian Wittmeier wrote: > On Wed, 5 May 2004 18:57:51 +0100, John Poltorak wrote: > > >If anyone is interested in trying to build Open Office can we agree an a > >release version? > > > >I suggest OOO_1_1_1:- > > > >http://ooo.ximian.com/packages/OOO_1_1_1/OOO_1_1_1.tar.bz2 > > cool, parts of the source still have German comments :-) You must know where to look, I guess. This archive is massive! Any signs of residual OS/2 support? What strikes me is the number of seperate open source apps included in the archive, a number of which, such as berkeleydb, dmake, expat, jpeg, zlib etc are already known to work on OS/2. The problem is knowing where to start... > Sebastian -- John **= Email 35 ==========================** Date: Thu, 06 May 2004 12:14:18 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: distcc John Poltorak schrieb: > distccd[12738] (main) Warning: nice 5 failed: Invalid > argument Sounds like decreasing priority doesn't work as expected, though that's not really critical. > distccd[12738] (dcc_discard_root) CRITICAL! still have > root privileges after trying to discard them! > distccd[12738] (dcc_exit) exit: code 112; self: 0.000000 > user 0.000000 sys; children: 0.000000 user 0.000000 sys > > > Looks like some dabbling with code may be required... Yes. On OS/2 you hardly can get rid of root priviledges, so this test needs to be disabled for OS/2 in the source code. HTH, Stefan **= Email 36 ==========================** Date: Thu, 06 May 2004 13:28:50 +0300 From: Sergey Yevtushenko Subject: Re: distcc Stefan, >> >>Looks like some dabbling with code may be required... > > Yes. On OS/2 you hardly can get rid of root > priviledges, so this test needs to be disabled > for OS/2 in the source code. Well, Security/2 can help to avoid root, but I agree that it's better to disable that "feature" in OS/2. I just think that setting up rights is not an application responsibility. Regards, Sergey. -- *--------------------------------------------- ES at Home http://es.os2.ru/ **= Email 37 ==========================** Date: Thu, 6 May 2004 11:34:33 +0100 From: John Poltorak Subject: Re: distcc On Thu, May 06, 2004 at 12:14:18PM +0200, Stefan.Neis at t-online.de wrote: > John Poltorak schrieb: > > > distccd[12738] (dcc_discard_root) CRITICAL! still have > > root privileges after trying to discard them! > > distccd[12738] (dcc_exit) exit: code 112; self: 0.000000 > > user 0.000000 sys; children: 0.000000 user 0.000000 sys > > > > > > Looks like some dabbling with code may be required... > > Yes. On OS/2 you hardly can get rid of root > priviledges, so this test needs to be disabled > for OS/2 in the source code. This is where I get out of my depth... AFAICT this is the code causing the problem:- (from distcc-2.14\src\setuid.c) int dcc_discard_root(void) { uid_t uid; gid_t gid; int ret; if (getuid() != 0 && geteuid() != 0) { /* Already not root. No worries. */ return 0; } if ((ret = dcc_preferred_user(&uid, &gid)) != 0) return ret; /* GNU C Library Manual says that when run by root, setgid() and setuid() * permanently discard privileges: both the real and effective uid are * set. */ if (setgid(gid)) { rs_log_error("setgid(%d) failed: %s", (int) gid, strerror(errno)); return EXIT_SETUID_FAILED; } #ifdef HAVE_SETGROUPS /* Get rid of any supplementary groups this process might have * inherited. */ /* XXX: OS X Jaguar broke setgroups so that setting it to 0 fails. */ if (setgroups(1, &gid)) { rs_log_error("setgroups failed: %s", strerror(errno)); return EXIT_SETUID_FAILED; } #endif if (setuid(uid)) { rs_log_error("setuid(%d) failed: %s", (int) uid, strerror(errno)); return EXIT_SETUID_FAILED; } if (getuid() == 0 || geteuid() == 0) { rs_log_crit("still have root privileges after trying to discard them!"); return EXIT_SETUID_FAILED; } rs_trace("discarded root privileges, changed to uid=%d gid=%d", (int) uid, (int) gid); return 0; } Any suggestions on where to apply some changes? > HTH, > Stefan -- John **= Email 38 ==========================** Date: Thu, 6 May 2004 11:39:26 +0100 From: John Poltorak Subject: Alternative to 'tar zxf' I've seen an alternative way of extracting from a .tgz file which didn't involve running 'tar zxf' but can't remember what it was. I think it involved piping the output from gzip to tar, but I can't stumble on the correct syntax. Can anyone remind me? -- John **= Email 39 ==========================** Date: Thu, 6 May 2004 07:14:23 -0400 (EDT) From: Thomas Dickey Subject: Re: Alternative to 'tar zxf' On Thu, 6 May 2004, John Poltorak wrote: > > > I've seen an alternative way of extracting from a .tgz file which didn't > involve running 'tar zxf' but can't remember what it was. I think it > involved piping the output from gzip to tar, but I can't stumble on the > correct syntax. > > Can anyone remind me? gzip -dc foo |tar xf - -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net **= Email 40 ==========================** Date: Thu, 6 May 2004 11:26:12 GMT From: Illya Vaes Subject: Re: Alternative to 'tar zxf' >I've seen an alternative way of extracting from a .tgz file which didn't >involve running 'tar zxf' but can't remember what it was. I think it >involved piping the output from gzip to tar, but I can't stumble on the >correct syntax. >Can anyone remind me? gzip -d -c | tar xvf - (or tvf - to show the contents) The '-' says "take from standard input" and '-c' to standard output) **= Email 41 ==========================** Date: Thu, 06 May 2004 13:55:38 +0200 (CEST) From: Stefan.Neis at t-online.de Subject: Re: distcc John Poltorak schrieb: > This is where I get out of my depth... Well, something like > int dcc_discard_root(void) > { #ifndef __EMX__ > uid_t uid; > gid_t gid; > int ret; > > if (getuid() != 0 && geteuid() != 0) { > /* Already not root. No worries. */ > return 0; > } > > if ((ret = dcc_preferred_user(&uid, &gid)) != 0) > return ret; > > /* GNU C Library Manual says that when run by root, > setgid() and setuid() > * permanently discard privileges: both the real and > effective uid are > * set. */ > > if (setgid(gid)) { > rs_log_error("setgid(%d) failed: %s", (int) gid, > strerror(errno)); > return EXIT_SETUID_FAILED; > } > > #ifdef HAVE_SETGROUPS > /* Get rid of any supplementary groups this process > might have > * inherited. */ > /* XXX: OS X Jaguar broke setgroups so that setting > it to 0 fails. */ > if (setgroups(1, &gid)) { > rs_log_error("setgroups failed: %s", > strerror(errno)); > return EXIT_SETUID_FAILED; > } > #endif > > if (setuid(uid)) { > rs_log_error("setuid(%d) failed: %s", (int) uid, > strerror(errno)); > return EXIT_SETUID_FAILED; > } > > if (getuid() == 0 || geteuid() == 0) { > rs_log_crit("still have root privileges after > trying to discard them!"); > return EXIT_SETUID_FAILED; > } > > rs_trace("discarded root privileges, changed to > uid=%d gid=%d", (int) uid, (int) gid); #endif > return 0; > } should do (Without the "> " at the beginning of lines, of course...) But whatever the result, I have some doubts that you really can get something generating code suitable for OS/2 from it. AFAIK, Innotek's changes have not been integrated into standard gcc distribution, or have they? Regards, Stefan **= Email 42 ==========================** Date: Thu, 06 May 2004 14:59:40 +0300 From: Sergey Yevtushenko Subject: Re: distcc John, > AFAICT this is the code causing the problem:- > (from distcc-2.14\src\setuid.c) > > int dcc_discard_root(void) > { > uid_t uid; > gid_t gid; > int ret; #ifdef __OS2___ return 0; #endif > if (getuid() != 0 && geteuid() != 0) { > /* Already not root. No worries. */ > return 0; > } > > if ((ret = dcc_preferred_user(&uid, &gid)) != 0) > return ret; > > /* GNU C Library Manual says that when run by root, setgid() and setuid() > * permanently discard privileges: both the real and effective uid are > * set. */ > > if (setgid(gid)) { > rs_log_error("setgid(%d) failed: %s", (int) gid, strerror(errno)); > return EXIT_SETUID_FAILED; > } > > #ifdef HAVE_SETGROUPS > /* Get rid of any supplementary groups this process might have > * inherited. */ > /* XXX: OS X Jaguar broke setgroups so that setting it to 0 fails. */ > if (setgroups(1, &gid)) { > rs_log_error("setgroups failed: %s", strerror(errno)); > return EXIT_SETUID_FAILED; > } > #endif > > if (setuid(uid)) { > rs_log_error("setuid(%d) failed: %s", (int) uid, strerror(errno)); > return EXIT_SETUID_FAILED; > } > > if (getuid() == 0 || geteuid() == 0) { > rs_log_crit("still have root privileges after trying to discard them!"); > return EXIT_SETUID_FAILED; > } > > rs_trace("discarded root privileges, changed to uid=%d gid=%d", (int) uid, (int) gid); > return 0; > } > > > Any suggestions on where to apply some changes? See above. Regards, Sergey. -- *--------------------------------------------- ES at Home http://es.os2.ru/ **= Email 43 ==========================** Date: Thu, 06 May 2004 08:36:15 -0400 From: "T.Sikora" Subject: Re: Alternative to 'tar zxf' John Poltorak wrote: > > I've seen an alternative way of extracting from a .tgz file which didn't > involve running 'tar zxf' but can't remember what it was. I think it > involved piping the output from gzip to tar, but I can't stumble on the > correct syntax. > > Can anyone remind me? > > This it? gzip -cd {%s} | tar tvf - -- T.Sikora tsikora at ntplx dot net