From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Mon, 25 Mar 2002 04:20:26 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 174 ************************************************** Sunday 24 March 2002 Number 174 ************************************************** Subjects for today 1 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : Jun Sawataishi 2 Re: IFS in shell scripts : Jun Sawataishi 3 Re: ncurses.a: record too long : Thomas E. Dickey" 4 Re: Re: Running ./configure : James Cannon 5 Re: ncurses.a: record too long : Thomas E. Dickey" 6 Re: ncurses.a: record too long : Marty" 7 ncurses.a: record too long : John Poltorak 8 Re: Re: Running ./configure : John Poltorak 9 Re: ncurses.a: record too long : Henry Sobotka 10 Re: Re: Running ./configure : Stefan Neis 11 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : Thomas E. Dickey" 12 Re: NCURSES headers : Thomas E. Dickey" 13 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : Thomas E. Dickey" 14 Re: ncurses.a: record too long : Henry Sobotka 15 Re: ncurses.a: record too long : John Poltorak 16 Re: giflib : John Poltorak 17 libgtop : John Poltorak 18 Re: ncurses.a: record too long : John Poltorak 19 Need some help builing GETTEXT : John Poltorak 20 Re: ncurses.a: record too long : John Poltorak 21 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : John Poltorak 22 NCURSES headers : John Poltorak 23 Re: ncurses.a: record too long : John Poltorak 24 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : John Poltorak 25 Re: AWK & SED : John Poltorak 26 Re: AWK & SED : Adrian Gschwend" 27 Re: Re: Building Perl.exe as a test of manhood ;-) : Stephen Amadei 28 Re: BIND : Tony Sanmarti" 29 Re: BIND : John Poltorak 30 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : Dave and Natalie" 31 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : John Poltorak 32 Re: Re: Running ./configure : Dave and Natalie" 33 Re: NCURSES headers : Dave and Natalie" 34 Re: ncurses.a: record too long : Marty" 35 Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 : Jun Sawataishi 36 Re: AWK & SED : John Poltorak 37 Re: AWK & SED : Adrian Gschwend" 38 Boost. Blah. : Stephen Amadei 39 Re: Boost. Blah. : Henry Sobotka 40 Re: ncurses.a: record too long : Jun Sawataishi 41 Re: Boost. Blah. : Stephen Amadei **= Email 1 ==========================** Date: Mon, 25 Mar 2002 00:04:58 +0900 From: Jun Sawataishi Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 At Sat, 23 Mar 2002 22:39:23 -0800, Dave and Natalie wrote: > On Sun, 24 Mar 2002 04:26:46 +0900, Jun Sawataishi wrote: > > - static link with ncurses 5.2 modified by me > http://homepage1.nifty.com/jsawa/gnu/ncurses-5.2.zip seems to be corrupt. It is 70k instead of 460k Sorry to trouble you. I just re-uploaded ncurses-5.2.zip. Please re-try again. Note that file size is changed from 460k to 261k. **= Email 2 ==========================** Date: Mon, 25 Mar 2002 00:38:32 +0900 From: Jun Sawataishi Subject: Re: IFS in shell scripts At Sun, 24 Mar 2002 12:43:42 +0000, John Poltorak wrote: > If I define IFS as a TAB does a single tab specify a field seperator? > ie if there are two tabs together are they interpreted as a single tab by > the shell? > I want to set up a line such as:- > p1TABp2TABp3TABp4 > How do I specify a nul value for p2? > What happens if I actually have? :- > abcTABTABxyzTAB123 > does p2 take the value xyz or nul ? Using sh, ksh, and bash I tested how IFS worked. From now on represents TAB charactor not "\t". #script for IFS: tabtest.sh IFS='' x='abc' for y in $x ; do echo -${y}_ done IFS=';' x='a;b;;c' for y in $x ; do echo -${y}_ done IFS='x' x='axbxxc' for y in $x ; do echo -${y}_ done Here I show the result. 1 [k:/tmp] 00:34:32<1003># sh tabtest.sh -a_ -b_ -c_ -a_ -b_ -_ <== NOTE that this represents a null string -c_ -a_ -b_ -_ <== NOTE that this represents a null string -c_ If you want to set a null string, do not use as IFS. # OS/2 is not a question, it's a solution. # SAWATAISHI Jun # http://www2s.biglobe.ne.jp/~vtgf3mpr/ **= Email 3 ==========================** Date: Mon, 25 Mar 2002 05:36:14 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: ncurses.a: record too long On Mon, 25 Mar 2002, John Poltorak wrote: > > When building NCURSES, I get this error:- > > > cd ../objects; gcc -I../ncurses -I. -DNDEBUG -I. -I../include > -I/usr/local/include -DHAVE_CONFIG_H -g -O -Zmt -D__ST_MT_ERRNO__ -c > ../ncurses/./tinfo/write_entry.c ... > emxomfar: ../lib/ncurses.a(../objects/hardscroll.o): Record too long > Creating library file `../lib/ncurses.a' > a - ../objects/hardscroll.o > make[1]: *** [../lib/ncurses.a] Error 2 > make[1]: Leaving directory `C:/unixos2/workdir/ncurses-5.2/ncurses' > make: *** [all] Error 2 > + make install > > > Can anyone shed any light on this? sometimes that's only a shell limitation, other times it's a system-level limit. -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 4 ==========================** Date: Mon, 25 Mar 2002 07:30:22 -0800 (PST) From: James Cannon Subject: Re: Re: Running ./configure Hi John, I'ld to volunteer to test out your Perl package. Hopefully, I can document my experiences with it and put on the 'ezine. Please let me know the steps required ... --- John Poltorak wrote: > On Sun, Mar 24, 2002 at 02:07:54PM +0100, Adrian > Gschwend wrote: > > On Sat, 23 Mar 2002 21:25:43 +0000, John Poltorak > wrote: > > > > Hi John, > > > > >We've come a long way in the last year, thanks to > great efforts by > > >yourself and a number of other porters. > > > > I would like to thank everyone on the list too, > especially you! > > Thanks, but I'm only trying to repackage the work > that other people are > providing. Not being a C programmer makes me feel > particularly hamstrung > as far as getting any porting done. > > > It's > > great to see that the unixos2 effort is heading > into the right > > direction. There is still a lot of work to do but > what you did with > > FHS-like packages and the effort for pushing > UNIXROOT is really great! > > Actually, I don't think Slackware is particulary FHS > compliant and I'd > really like to start rebuilding all the packages > fairly soon from scratch > using a better directory structure. So far I have > been sticking slavishly > to Slackware's structure which means that certain > packages, such as the > GNU Text Utils (TXTUTILS) have executables split > between /bin and > /usr/bin. I don't particularly like this split and > it looks fairly > arbitrary to me. > > > > I hope we can improve again this year, it looks > good so far! I just > > hope we find a good way to clone Holger and some > other members on this > > list soon ;-))) > > I think that once the correct build toolset is > defined and available > people won't face the same frustrations as some of > us have experienced > over the last year. It's only by more people trying > things out and > relaying any problems that we can hope to refine the > final product. > > So I'd urge anyone who hasn't already done so to > build a package on OS/2. > Maybe try building Perl - it's a very useful > exercise, and quite > satisfying when you come up with a final runnable > PERL.EXE. > > > > > cu > > > > Adrian > > > > > > > > -- > > Adrian Gschwend > > at OS/2 Netlabs > > > > -- > John > > ===== Sincerely, James Cannon Using OS/2 Warp in the beautiful Wine Country of Northen California! __________________________________________________ Do You Yahoo!? Yahoo! Movies - coverage of the 74th Academy Awards® http://movies.yahoo.com/ **= Email 5 ==========================** Date: Mon, 25 Mar 2002 08:24:53 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: ncurses.a: record too long On Mon, 25 Mar 2002, John Poltorak wrote: > > sometimes that's only a shell limitation, other times it's a system-level > > limit. > > What are the implications of the error? Does it indicate a bad build? Is > there a work round? the usual work-around is to split the list into pieces. Is this for 5.2, or for one of the more recent snapshots? (5.2 built fine for me). > > The shell used was PDKSH - BASH doesn't seem to work very well with > NCURSES. -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 6 ==========================** Date: Mon, 25 Mar 2002 09:09:59 -0500 (EST) From: "Marty" Subject: Re: ncurses.a: record too long On Mon, 25 Mar 2002 10:08:48 +0000, John Poltorak wrote: > Can anyone shed any light on this? This has nothing to do with shell limitations. It has to do with the OMF page size. Try using -p4096 or a higher number. I ran into this problem the first time I tried to build MAME. **= Email 7 ==========================** Date: Mon, 25 Mar 2002 10:08:48 +0000 From: John Poltorak Subject: ncurses.a: record too long When building NCURSES, I get this error:- cd ../objects; gcc -I../ncurses -I. -DNDEBUG -I. -I../include -I/usr/local/include -DHAVE_CONFIG_H -g -O -Zmt -D__ST_MT_ERRNO__ -c ../ncurses/./tinfo/write_entry.c emxomfar rv ../lib/ncurses.a ../objects/hardscroll.o ../objects/hashmap.o ../objects/lib_addch.o ../objects/lib_addstr.o ../objects/lib_beep.o ../objects/lib_bkgd.o ../objects/lib_box.o ../objects/lib_chgat.o ../objects/lib_clear.o ../objects/lib_clearok.o ../objects/lib_clrbot.o ../objects/lib_clreol.o ../objects/lib_color.o ../objects/lib_colorset.o ../objects/lib_delch.o ../objects/lib_delwin.o ../objects/lib_echo.o ../objects/lib_endwin.o ../objects/lib_erase.o ../objects/lib_flash.o ../objects/lib_gen.o ../objects/lib_getch.o ../objects/lib_getstr.o ../objects/lib_hline.o ../objects/lib_immedok.o ../objects/lib_inchstr.o ../objects/lib_initscr.o ../objects/lib_insch.o ../objects/lib_insdel.o ../objects/lib_insstr.o ../objects/lib_instr.o ../objects/lib_isendwin.o ../objects/lib_leaveok.o ../objects/lib_mouse.o ../objects/lib_move.o ../objects/lib_mvcur.o ../objects/lib_mvwin.o ../objects/lib_newterm.o ../objects/lib_newwin.o ../objects/lib_nl.o ../objects/lib_overlay.o ../objects/lib_pad.o ../objects/lib_printw.o ../objects/lib_redrawln.o ../objects/lib_refresh.o ../objects/lib_restart.o ../objects/lib_scanw.o ../objects/lib_screen.o ../objects/lib_scroll.o ../objects/lib_scrollok.o ../objects/lib_scrreg.o ../objects/lib_set_term.o ../objects/lib_slk.o ../objects/lib_slkatr_set.o ../objects/lib_slkatrof.o ../objects/lib_slkatron.o ../objects/lib_slkatrset.o ../objects/lib_slkattr.o ../objects/lib_slkclear.o ../objects/lib_slkcolor.o ../objects/lib_slkinit.o ../objects/lib_slklab.o ../objects/lib_slkrefr.o ../objects/lib_slkset.o ../objects/lib_slktouch.o ../objects/lib_touch.o ../objects/lib_tstp.o ../objects/lib_ungetch.o ../objects/lib_vidattr.o ../objects/lib_vline.o ../objects/lib_wattroff.o ../objects/lib_wattron.o ../objects/lib_winch.o ../objects/lib_window.o ../objects/nc_panel.o ../objects/safe_sprintf.o ../objects/tty_update.o ../objects/memmove.o ../objects/sigaction.o ../objects/vsscanf.o ../objects/define_key.o ../objects/expanded.o ../objects/keybound.o ../objects/keyok.o ../objects/lib_dft_fgbg.o ../objects/lib_freeall.o ../objects/lib_print.o ../objects/resizeterm.o ../objects/tries.o ../objects/version.o ../objects/wresize.o ../objects/access.o ../objects/add_tries.o ../objects/alloc_entry.o ../objects/alloc_ttype.o ../objects/captoinfo.o ../objects/codes.o ../objects/comp_captab.o ../objects/comp_error.o ../objects/comp_expand.o ../objects/comp_hash.o ../objects/comp_parse.o ../objects/comp_scan.o ../objects/doalloc.o ../objects/fallback.o ../objects/free_ttype.o ../objects/getenv_num.o ../objects/home_terminfo.o ../objects/init_keytry.o ../objects/lib_acs.o ../objects/lib_baudrate.o ../objects/lib_cur_term.o ../objects/lib_data.o ../objects/lib_has_cap.o ../objects/lib_kernel.o ../objects/lib_keyname.o ../objects/lib_longname.o ../objects/lib_napms.o ../objects/lib_options.o ../objects/lib_raw.o ../objects/lib_setup.o ../objects/lib_termcap.o ../objects/lib_termname.o ../objects/lib_tgoto.o ../objects/lib_ti.o ../objects/lib_tparm.o ../objects/lib_tputs.o ../objects/lib_trace.o ../objects/lib_ttyflags.o ../objects/lib_twait.o ../objects/name_match.o ../objects/names.o ../objects/parse_entry.o ../objects/read_entry.o ../objects/read_termcap.o ../objects/setbuf.o ../objects/strings.o ../objects/unctrl.o ../objects/write_entry.o emxomfar: ../lib/ncurses.a(../objects/hardscroll.o): Record too long Creating library file `../lib/ncurses.a' a - ../objects/hardscroll.o make[1]: *** [../lib/ncurses.a] Error 2 make[1]: Leaving directory `C:/unixos2/workdir/ncurses-5.2/ncurses' make: *** [all] Error 2 + make install Can anyone shed any light on this? -- John **= Email 8 ==========================** Date: Mon, 25 Mar 2002 10:16:33 +0000 From: John Poltorak Subject: Re: Re: Running ./configure On Mon, Mar 25, 2002 at 10:25:52AM +0100, Stefan Neis wrote: > Hi, > > > >(and please don't tell me that I must never use a drive > > >letter because a) it's not Posix and b) it's much more > > >fun if the program behaves different on different drives) > > Yes, it's much more fun, if it behaves different on different > machines. :-( > If drive letters are compiled into the code, this essentially > means distributing precompiled binaries is completely useless > as they won't work anywhere else (Over here, e.g. C: is > either a FAT or an NTFS partition, depending on the version of > the DOS which I booted the last time). > So it remains to say: > - either don't use drive letters and wait for libemu to take > care of it automatically at some later point of time, i.e. the > programs behave different on different drives _for_now_ - but > no longer once they are linked against libemu instead of EMX - > but they are minimally functional _for_everyone_. > - or if you are willing to spend some more time on it, query the > environment variable UNIXROOT and prepend that to the path. > This will especially add a (configurable) drive letter. It's a > bit frustrating that this code can just be removed a bit later, > but for now it seems to be the best solution (at least to me). IMV, the best solution is for installers is to rebuild the app, tailored for its local environment by running:- Autoconf Configure Make Make install This isn't recommended just yet, but I think there may be a good chance of success in a couple of months. > Regards, > Stefan -- John **= Email 9 ==========================** Date: Mon, 25 Mar 2002 10:25:47 -0500 From: Henry Sobotka Subject: Re: ncurses.a: record too long John Poltorak wrote: > > When building NCURSES, I get this error:- > > cd ../objects; gcc -I../ncurses -I. -DNDEBUG -I. -I../include > -I/usr/local/include -DHAVE_CONFIG_H -g -O -Zmt -D__ST_MT_ERRNO__ -c > ../ncurses/./tinfo/write_entry.c > emxomfar rv ../lib/ncurses.a ../objects/hardscroll.o ../objects/hashmap.o [snip] > emxomfar: ../lib/ncurses.a(../objects/hardscroll.o): Record too long > Creating library file `../lib/ncurses.a' > a - ../objects/hardscroll.o > make[1]: *** [../lib/ncurses.a] Error 2 > make[1]: Leaving directory `C:/unixos2/workdir/ncurses-5.2/ncurses' > make: *** [all] Error 2 > + make install > > Can anyone shed any light on this? You're mixing formats again. Try adding -Zomf to your compilation flags or using ar instead of emxomfar. h~ **= Email 10 ==========================** Date: Mon, 25 Mar 2002 10:25:52 +0100 (CET) From: Stefan Neis Subject: Re: Re: Running ./configure Hi, > >(and please don't tell me that I must never use a drive > >letter because a) it's not Posix and b) it's much more > >fun if the program behaves different on different drives) Yes, it's much more fun, if it behaves different on different machines. :-( If drive letters are compiled into the code, this essentially means distributing precompiled binaries is completely useless as they won't work anywhere else (Over here, e.g. C: is either a FAT or an NTFS partition, depending on the version of the DOS which I booted the last time). So it remains to say: - either don't use drive letters and wait for libemu to take care of it automatically at some later point of time, i.e. the programs behave different on different drives _for_now_ - but no longer once they are linked against libemu instead of EMX - but they are minimally functional _for_everyone_. - or if you are willing to spend some more time on it, query the environment variable UNIXROOT and prepend that to the path. This will especially add a (configurable) drive letter. It's a bit frustrating that this code can just be removed a bit later, but for now it seems to be the best solution (at least to me). Regards, Stefan -- Micro$oft is not an answer. It is a question. The answer is 'no'. **= Email 11 ==========================** Date: Mon, 25 Mar 2002 10:31:44 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Mon, 25 Mar 2002, John Poltorak wrote: > Is this the same program as the one included in LESS? :- > > http://www.greenwoodsoftware.com/less/less-374.tar.gz preferably not. (it would make ncurses unusable in a number of cases). > > b) availability of Japanese DBCS in menu > > Without this modification, Japanese SJIS character directory name > > cannot be shown in file name completion. > > > Is there any way this could be included in the main NCURSES archive? The current development version of ncurses can be configured to use wide characters (which covers this). -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 12 ==========================** Date: Mon, 25 Mar 2002 12:17:00 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: NCURSES headers On Mon, 25 Mar 2002, John Poltorak wrote: > > > Is there as standard location for NCURSES headers? > > /usr/include > > or > > /usr/include/ncurses either of the above. > If the latter, then I guess we also need:- ... > > /usr/include/curses.h only if you have only one version of curses installed. I typically run on systems where this is not the case. -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 13 ==========================** Date: Mon, 25 Mar 2002 12:18:04 -0500 (EST) From: "Thomas E. Dickey" Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Mon, 25 Mar 2002, John Poltorak wrote: > On Mon, Mar 25, 2002 at 10:31:44AM -0500, Thomas E. Dickey wrote: > > On Mon, 25 Mar 2002, John Poltorak wrote: > > > > > Is this the same program as the one included in LESS? :- > > > > > > http://www.greenwoodsoftware.com/less/less-374.tar.gz > > > > preferably not. (it would make ncurses unusable in a number of cases). > > What's the problem with SCRSIZE? It's a program included with LESS so many > people are likely to have it already. it links to the X library. -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net **= Email 14 ==========================** Date: Mon, 25 Mar 2002 12:20:35 -0500 From: Henry Sobotka Subject: Re: ncurses.a: record too long John Poltorak wrote: > > I'm trying to create a single build script for as many apps as possible > and was using:- > > make AR=emxomfar > > which I just changed to:- > > make > > and it works now. Is AR=emxomfar the sort of thing that $MAKEFLAGS is > supposed to deal with? Whether or not AR should be set to emxomfar depends on whether or not CFLAGS includes -Zomf. OMF is preferable for DLLs. Probably the simplest way to handle it universally is to do an aout build and then feed the static .a lib to Andy's dllar.cmd. h~ **= Email 15 ==========================** Date: Mon, 25 Mar 2002 12:47:54 +0000 From: John Poltorak Subject: Re: ncurses.a: record too long On Mon, Mar 25, 2002 at 05:36:14AM -0500, Thomas E. Dickey wrote: > On Mon, 25 Mar 2002, John Poltorak wrote: > > > > > When building NCURSES, I get this error:- > > emxomfar: ../lib/ncurses.a(../objects/hardscroll.o): Record too long > > Creating library file `../lib/ncurses.a' > > a - ../objects/hardscroll.o > > make[1]: *** [../lib/ncurses.a] Error 2 > > make[1]: Leaving directory `C:/unixos2/workdir/ncurses-5.2/ncurses' > > make: *** [all] Error 2 > > + make install > > > > > > Can anyone shed any light on this? > > sometimes that's only a shell limitation, other times it's a system-level > limit. What are the implications of the error? Does it indicate a bad build? Is there a work round? The shell used was PDKSH - BASH doesn't seem to work very well with NCURSES. > -- > T.E.Dickey -- John **= Email 16 ==========================** Date: Mon, 25 Mar 2002 13:13:41 +0000 From: John Poltorak Subject: Re: giflib On Sat, Mar 23, 2002 at 08:07:23PM -0800, mlaitio wrote: > Maybe they have removed it after somebody claimed to have patent from > gif and started > requiring money... That was at least the main driving force why people > started to design > png-format. GIFs are still very common and I would be very surprised if Slackware did not contain some libraries which support GIF. Maybe this is what I'm looking for:- .... ftp://ftp.mirror.ac.uk/sites/ftp.slackware.com/pub/slackware/slackware/slakware/gtk1/libungif.tgz > Mika > > John Poltorak wrote: > > >On Sat, Mar 09, 2002 at 12:09:44PM +0000, Dave and Natalie wrote: > > > >>Has anyone ported giflib (ftp://sunsite.unc.edu/pub/Linux/libs/giflib/giflib-4.1.0.tar.gz) ideally with dll lib and headers? > >> > > > >GIFLIB sounds like the sort of thing we ought to have have in UnixOS/2 > >since we already have JPEG6, LIBPNG and LIBTIFF which are based on the > >Slackware packages, but there doesn't seem to be a package for GIF files > >in the distro. > > > >If anyone is familiar with Slackware can you see a package for providing > >GIF support? > > > > > >>Dave -- John **= Email 17 ==========================** Date: Mon, 25 Mar 2002 13:19:40 +0000 From: John Poltorak Subject: libgtop Whilst looking through the contents of Slackware's packages, I came across libgtop (in GTK1 series) :- libgtop-1.0.12 A library that fetches information about the running system such as cpu and memory usage, active processes etc. On Linux systems, these information are taken directly from the /proc filesystem. Do we have anything like this available for OS/2? -- John **= Email 18 ==========================** Date: Mon, 25 Mar 2002 13:49:10 +0000 From: John Poltorak Subject: Re: ncurses.a: record too long On Mon, Mar 25, 2002 at 08:24:53AM -0500, Thomas E. Dickey wrote: > On Mon, 25 Mar 2002, John Poltorak wrote: > > > > sometimes that's only a shell limitation, other times it's a system-level > > > limit. > > > > What are the implications of the error? Does it indicate a bad build? Is > > there a work round? > > the usual work-around is to split the list into pieces. Is this for 5.2, > or for one of the more recent snapshots? (5.2 built fine for me). It's this one:- ftp://ftp.gnu.org/pub/gnu/ncurses/ncurses-5.2.tar.gz I'm attempting to build it in a new environment, so it may be that I have missed something out in setting it up. I do get a number of .a files installed in /usr/local/lib including ncurses.a, but I'm just trying to iron out any error msgs so that I get as good an OS/2 build as I can hope for. Do you have a config.log for your build so I can compare notes? > -- > T.E.Dickey -- John **= Email 19 ==========================** Date: Mon, 25 Mar 2002 14:12:41 +0000 From: John Poltorak Subject: Need some help builing GETTEXT If anyone has managed to build this:- ftp://ftp.gnu.org/pub/gnu/gettext/gettext-0.11.1.tar.gz I'd appreciate a few tips.... -- John **= Email 20 ==========================** Date: Mon, 25 Mar 2002 14:16:54 +0000 From: John Poltorak Subject: Re: ncurses.a: record too long On Mon, Mar 25, 2002 at 09:09:59AM -0500, Marty wrote: > On Mon, 25 Mar 2002 10:08:48 +0000, John Poltorak wrote: > > > Can anyone shed any light on this? > > This has nothing to do with shell limitations. It has to do with the OMF page size. > Try using -p4096 or a higher number. I ran into this problem the first time I tried > to build MAME. Where do I specify this? Is it possible to make this a default setting, and if so will it cause any problems/overhead for other apps? I'm trying to devise a generic build script so would prefer as few exceptions as possible. -- John **= Email 21 ==========================** Date: Mon, 25 Mar 2002 14:23:43 +0000 From: John Poltorak Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Mon, Mar 25, 2002 at 09:01:05PM +0900, Jun Sawataishi wrote: > At Sun, 24 Mar 2002 16:44:15 +0000, > John Poltorak wrote: > > How does your port of NCURSES differ from the real thing? > > a) can determine correct window size in XFree86 virtual terminal, using > scrsize.exe. > `scrsize.c' is a freeware written by Mr. Altair, a Japanese OS/2 > hacker. Is this the same program as the one included in LESS? :- http://www.greenwoodsoftware.com/less/less-374.tar.gz > b) availability of Japanese DBCS in menu > Without this modification, Japanese SJIS character directory name > cannot be shown in file name completion. Is there any way this could be included in the main NCURSES archive? It would be nice to use a single distro, especially when that distro explicitly contains OS/2 support as is the case with NCURSES. > # SAWATAISHI Jun -- John **= Email 22 ==========================** Date: Mon, 25 Mar 2002 16:19:07 +0000 From: John Poltorak Subject: NCURSES headers Is there as standard location for NCURSES headers? /usr/include or /usr/include/ncurses If the latter, then I guess we also need:- ... /usr/include/curses.h which consists of:- #include Is there a configure or install option for automatically creating this file? -- John **= Email 23 ==========================** Date: Mon, 25 Mar 2002 16:50:13 +0000 From: John Poltorak Subject: Re: ncurses.a: record too long On Mon, Mar 25, 2002 at 10:25:47AM -0500, Henry Sobotka wrote: > John Poltorak wrote: > > > > When building NCURSES, I get this error:- > > > [snip] > > > emxomfar: ../lib/ncurses.a(../objects/hardscroll.o): Record too long > > Creating library file `../lib/ncurses.a' > > > > Can anyone shed any light on this? > > You're mixing formats again. Yes, you're right. > Try adding -Zomf to your compilation flags > or using ar instead of emxomfar. I'm trying to create a single build script for as many apps as possible and was using:- make AR=emxomfar which I just changed to:- make and it works now. Is AR=emxomfar the sort of thing that $MAKEFLAGS is supposed to deal with? If so, I guess I can add it to my parameter table... > h~ -- John **= Email 24 ==========================** Date: Mon, 25 Mar 2002 16:55:53 +0000 From: John Poltorak Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Mon, Mar 25, 2002 at 10:31:44AM -0500, Thomas E. Dickey wrote: > On Mon, 25 Mar 2002, John Poltorak wrote: > > > Is this the same program as the one included in LESS? :- > > > > http://www.greenwoodsoftware.com/less/less-374.tar.gz > > preferably not. (it would make ncurses unusable in a number of cases). What's the problem with SCRSIZE? It's a program included with LESS so many people are likely to have it already. > -- > T.E.Dickey -- John **= Email 25 ==========================** Date: Mon, 25 Mar 2002 17:17:00 +0000 From: John Poltorak Subject: Re: AWK & SED On Mon, Mar 25, 2002 at 06:03:31PM +0100, Adrian Gschwend wrote: > On Sat, 23 Mar 2002 10:43:17 +0000, John Poltorak wrote: > > Hi John, > > >It's a very important package, one that everyone should try out. > > I tried awk and I got this: > > SYS1804: The system cannot find the file REGEX., same goes for SED. I > have: > > Directory of D:\unixos2\usr\lib > > 17.06.99 0.40 29993 103 regex012.dll > > is that the one needed or is there something missing? You need the OS2LIBS package from the A1 series. The OS2LIBS package is meant to be a bundle of common DLLs shared by a number of applications. At the moment, all it contains is INTL and REGEX as well as EMXRT. If anyone can think of any other candidates to be included that would be useful. I would like to see single versions of these two DLLs instead of the many variations that we have at the moment. > cu > > Adrian > > -- John **= Email 26 ==========================** Date: Mon, 25 Mar 2002 18:03:31 +0100 (CET) From: "Adrian Gschwend" Subject: Re: AWK & SED On Sat, 23 Mar 2002 10:43:17 +0000, John Poltorak wrote: Hi John, >It's a very important package, one that everyone should try out. I tried awk and I got this: SYS1804: The system cannot find the file REGEX., same goes for SED. I have: Directory of D:\unixos2\usr\lib 17.06.99 0.40 29993 103 regex012.dll is that the one needed or is there something missing? cu Adrian -- Adrian Gschwend at OS/2 Netlabs ICQ: 22419590 ktk at netlabs.org ------- The OS/2 OpenSource Project: http://www.netlabs.org **= Email 27 ==========================** Date: Mon, 25 Mar 2002 18:38:49 -0500 (EST) From: Stephen Amadei Subject: Re: Re: Building Perl.exe as a test of manhood ;-) On Sun, 24 Mar 2002, John Poltorak wrote: > So I'd urge anyone who hasn't already done so to build a package on OS/2. > Maybe try building Perl - it's a very useful exercise, and quite > satisfying when you come up with a final runnable PERL.EXE. IMHO, building Perl is a twisted intro to building packages on OS/2. It's been three years, and I am _still_ unable to get past DynaLoader in the Perl build (so I could get mod_perl to run). Of course, since the Perl build is not only C, it is my weakness. Normally, when building a package, if I loose my way, I just start compiling things one at a time. ----Steve Stephen Amadei Dandy.NET! CTO Atlantic City, NJ **= Email 28 ==========================** Date: Mon, 25 Mar 2002 18:52:33 +0100 (CET) From: "Tony Sanmarti" Subject: Re: BIND >AFAIK the porter was intending to work on v9 at some point but I haven't >heard from him for a while. Well, I'm still alive :-) However, I really don't have enougt time to keep on working on Bind :-( Tony. -------------------------------------------------------------------------------- Using OS/2 because I want to, not because I "have to". **= Email 29 ==========================** Date: Mon, 25 Mar 2002 19:01:50 +0000 From: John Poltorak Subject: Re: BIND On Mon, Mar 25, 2002 at 06:52:33PM +0100, Tony Sanmarti wrote: > >AFAIK the porter was intending to work on v9 at some point but I haven't > >heard from him for a while. > > Well, I'm still alive :-) > However, I really don't have enougt time to keep on working on Bind :-( That's fair enough. You've done a good job getting it updated. I take it that you are wanting to hand it on to someone else... > > Tony. -- John **= Email 30 ==========================** Date: Mon, 25 Mar 2002 19:58:57 -0800 From: "Dave and Natalie" Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Mon, 25 Mar 2002 12:18:04 -0500 (EST), Thomas E. Dickey wrote: >> What's the problem with SCRSIZE? It's a program included with LESS so many >> people are likely to have it already. > >it links to the X library. Can it not be rebuilt with static linking? Dave **= Email 31 ==========================** Date: Mon, 25 Mar 2002 20:00:28 +0000 From: John Poltorak Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 On Tue, Mar 26, 2002 at 02:19:14AM +0900, Jun Sawataishi wrote: > At Mon, 25 Mar 2002 16:55:53 +0000, > John Poltorak wrote: > > > > > > > Is this the same program as the one included in LESS? :- > Yes. > > > > http://www.greenwoodsoftware.com/less/less-374.tar.gz > > > preferably not. (it would make ncurses unusable in a number of cases). > > What's the problem with SCRSIZE? It's a program included with LESS so many > > people are likely to have it already. > scrsize.exe requires X11.dll, because it is linked with X11.a, not X11_s.a. > So, scrsize.exe fails if X11.dll is not available. I don't know who provided the OS/2 Makefile for LESS, but it includes this line:- scrsize.exe: scrsize.c ${CC} ${CFLAGS} -D__ST_MT_ERRNO__ -s -Zmtd -lX11 $< Does this mean that if I have the correct X11 libraries it will build and work properly? BTW I don't see any reference to X11 in my copy of scrsize.... > # SAWATAISHI Jun -- John **= Email 32 ==========================** Date: Mon, 25 Mar 2002 20:01:37 -0800 From: "Dave and Natalie" Subject: Re: Re: Running ./configure On Mon, 25 Mar 2002 07:30:22 -0800 (PST), James Cannon wrote: >Hi John, > >I'ld to volunteer to test out your Perl package. >Hopefully, I can document my experiences with it and >put on the 'ezine. Please let me know the steps >required ... I'll echo this, back to building gettext Dave **= Email 33 ==========================** Date: Mon, 25 Mar 2002 20:11:49 -0800 From: "Dave and Natalie" Subject: Re: NCURSES headers On Mon, 25 Mar 2002 12:17:00 -0500 (EST), Thomas E. Dickey wrote: >On Mon, 25 Mar 2002, John Poltorak wrote: > >> >> >> Is there as standard location for NCURSES headers? >> >> /usr/include >> >> or >> >> /usr/include/ncurses > >either of the above. > >> If the latter, then I guess we also need:- ... >> >> /usr/include/curses.h > >only if you have only one version of curses installed. I typically run >on systems where this is not the case. Debian has an elegant way of handling this. There is an /etc/alternatives directory where all programs that have multiple choices are symlinked. eg pager could be linked against more or less, a better example is windowmanager, where x-window-manager is linked against whatever windowmanager you want to use. This way you only have to change one symlink (often with a script) to change your windowmanager. Linux uses symlinks a lot to handle multiple versions or different programs to do a standard thing. eg curses.h can be linked against ncurses.h or whatever. Dave **= Email 34 ==========================** Date: Mon, 25 Mar 2002 20:31:41 -0500 (EST) From: "Marty" Subject: Re: ncurses.a: record too long On Mon, 25 Mar 2002 14:16:54 +0000, John Poltorak wrote: > On Mon, Mar 25, 2002 at 09:09:59AM -0500, Marty wrote: > > On Mon, 25 Mar 2002 10:08:48 +0000, John Poltorak wrote: > > > > > Can anyone shed any light on this? > > > > This has nothing to do with shell limitations. It has to do with the > > OMF page size. Try using -p4096 or a higher number. I ran into this > > problem the first time I tried to build MAME. > > Where do I specify this? On either the emxomf or emxomfar command line. > Is it possible to make this a default setting, and if so will it cause any > problems/overhead for other apps? I believe it will make your libraries/object files bigger (but not by much). It makes the allocation size more granular I believe. I needed to do this for some CPU emulation cores, which were huge, when I attempted to stuff them in an OMF library. On Mon, 25 Mar 2002 10:25:47 -0500, Henry Sobotka wrote: > > You're mixing formats again. Try adding -Zomf to your compilation flags > or using ar instead of emxomfar. I thought emxomfar handled the conversion as it stuffed them in the library, but I might be wrong. It's been a while. I do know that ar librarians will take any old file you give them and attempt to stuff them in the library whether they're valid modules or not, so it's true that it wouldn't complain about mixing formats. **= Email 35 ==========================** Date: Mon, 25 Mar 2002 21:01:05 +0900 From: Jun Sawataishi Subject: Re: New binary package for GNU Bash 2.05a and GNU Texinfo 4.1 At Sun, 24 Mar 2002 16:44:15 +0000, John Poltorak wrote: > How does your port of NCURSES differ from the real thing? a) can determine correct window size in XFree86 virtual terminal, using scrsize.exe. `scrsize.c' is a freeware written by Mr. Altair, a Japanese OS/2 hacker. b) availability of Japanese DBCS in menu Without this modification, Japanese SJIS character directory name cannot be shown in file name completion. # OS/2 is not a question, it's a solution. # SAWATAISHI Jun # http://www2s.biglobe.ne.jp/~vtgf3mpr/ **= Email 36 ==========================** Date: Mon, 25 Mar 2002 21:03:20 +0000 From: John Poltorak Subject: Re: AWK & SED On Mon, Mar 25, 2002 at 09:16:12PM +0100, Adrian Gschwend wrote: > On Mon, 25 Mar 2002 17:17:00 +0000, John Poltorak wrote: > > >You need the OS2LIBS package from the A1 series. > > thanks, I try to fetch that but I always get that too many users are > online (just 5 according to the message!?). Bad luck tonight? Yes, I get that all the time. Wonder why it is such a low limit... If you are desperate you could try:- ftp://www.eyup.org/pkg/os2libs.zip > >The OS2LIBS package is meant to be a bundle of common DLLs shared by a > >number of applications. At the moment, all it contains is INTL and REGEX > >as well as EMXRT. > > so I don't need libemx anymore? Probably not. > cu > > Adrian -- John **= Email 37 ==========================** Date: Mon, 25 Mar 2002 21:16:12 +0100 (CET) From: "Adrian Gschwend" Subject: Re: AWK & SED On Mon, 25 Mar 2002 17:17:00 +0000, John Poltorak wrote: >You need the OS2LIBS package from the A1 series. thanks, I try to fetch that but I always get that too many users are online (just 5 according to the message!?). Bad luck tonight? >The OS2LIBS package is meant to be a bundle of common DLLs shared by a >number of applications. At the moment, all it contains is INTL and REGEX >as well as EMXRT. so I don't need libemx anymore? cu Adrian -- Adrian Gschwend at OS/2 Netlabs ICQ: 22419590 ktk at netlabs.org ------- The OS/2 OpenSource Project: http://www.netlabs.org **= Email 38 ==========================** Date: Mon, 25 Mar 2002 22:27:32 -0500 (EST) From: Stephen Amadei Subject: Boost. Blah. O.K... my fighting with ./configure has taken a back seat, as the laterest error is... "Boost not found. Install Boost libraries. www.boost.org" Well, I have downloaded this thing, and it's got some kind of "portible" make system called Jam. Well, I managed to get Jam to compile in EMX, and it appears to work pretty well. Out of 712 targets needed for creation/compilation 688 worked. about 24 die because g++ is being called. I have always compiled c++ with gcc... what exact is the difference between gcc and g++. I made a copy of gcc into g++, but it won't compile... something about "sorry not implemented: namespace". Is there a way around this? I guess I'll look on Hobbes for a g++. ----Steve Stephen Amadei Dandy.NET! CTO Atlantic City, NJ **= Email 39 ==========================** Date: Mon, 25 Mar 2002 22:57:58 -0500 From: Henry Sobotka Subject: Re: Boost. Blah. Stephen Amadei wrote: > > Out of 712 targets needed for creation/compilation 688 worked. about 24 > die because g++ is being called. I have always compiled c++ with gcc... > what exact is the difference between gcc and g++. I made a copy of gcc > into g++, but it won't compile... something about "sorry not implemented: > namespace". > > Is there a way around this? I guess I'll look on Hobbes for a g++. You'll find a g++.exe with pgcc and gcc 3.x. With 2.8.1 and pgcc, gcc is a frontend for C++ compilation, but with 3.x you have to invoke g++ directly. Don't know the whys and wherefors. In any case, pgcc and 3.x, are far more advanced for C++ than 2.8.1. h~ **= Email 40 ==========================** Date: Mon, 25 Mar 2002 23:13:42 +0900 From: Jun Sawataishi Subject: Re: ncurses.a: record too long At Mon, 25 Mar 2002 10:08:48 +0000, John Poltorak wrote: > When building NCURSES, I get this error:- > cd ../objects; gcc -I../ncurses -I. -DNDEBUG -I. -I../include > -I/usr/local/include -DHAVE_CONFIG_H -g -O -Zmt -D__ST_MT_ERRNO__ -c > ../ncurses/./tinfo/write_entry.c > emxomfar rv ../lib/ncurses.a ../objects/hardscroll.o ../objects/hashmap.o > ../objects/lib_addch.o ../objects/lib_addstr.o ../objects/lib_beep.o > ../objects/lib_bkgd.o ../objects/lib_box.o ../objects/lib_chgat.o > ../objects/lib_clear.o ../objects/lib_clearok.o ../objects/lib_clrbot.o .................. > Can anyone shed any light on this? When you use emxomfar, object files must be compiled with "-Zomf". # OS/2 is not a question, it's a solution. # SAWATAISHI Jun # http://www2s.biglobe.ne.jp/~vtgf3mpr/ **= Email 41 ==========================** Date: Mon, 25 Mar 2002 23:32:32 -0500 (EST) From: Stephen Amadei Subject: Re: Boost. Blah. On Mon, 25 Mar 2002, Henry Sobotka wrote: > > Is there a way around this? I guess I'll look on Hobbes for a g++. > > You'll find a g++.exe with pgcc and gcc 3.x. With 2.8.1 and pgcc, gcc is > a frontend for C++ compilation, but with 3.x you have to invoke g++ > directly. Don't know the whys and wherefors. In any case, pgcc and 3.x, > are far more advanced for C++ than 2.8.1. The only pgcc I see is 1.1.1, which on the webpage link is gcc 2.95.3. Where can I find a gcc 3.x for OS/2? Thanx in advance. ----Steve Stephen Amadei Dandy.NET! CTO Atlantic City, NJ