Date: Wed, 10 Sep 2003 02:49:37 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [Ux2bs_Archive] No. 198 ************************************************** Tuesday 09 September 2003 Number 198 ************************************************** Subjects for today 1 Re: Building db (sleepycat) : James Moe" 2 gcc2 : T.Sikora" 3 Re: Building db (sleepycat) : Stefan.Neis at t-online.de 4 Re: Building db (sleepycat) : James Moe" 5 Re: Building db (sleepycat) : Stefan.Neis at t-online.de 6 Possible problem with read() and write()? : James Moe" 7 Possible problem with read() and write()? -- Never mind : James Moe" 8 Re: Possible problem with read() and write()? : Henry Sobotka 9 Re: Possible problem with read() and write()? -- Never mind : Henry Sobotka **= Email 1 ==========================** Date: Wed, 10 Sep 2003 08:33:59 -0700 (MST) From: "James Moe" Subject: Re: Building db (sleepycat) -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Wed, 10 Sep 2003 12:42:06 +0100, Stefan.Neis at t-online.de wrote: >The problem might be that some shell code abuses IFS to "tokenize" >some arbitrary strings, ... > Yes, this is exactly is happening. However the point is this: The assignment IFS=: is IGNORED, NOT DONE, BYPASSED, BLOWN OFF, SNUBBED, NOT HAPPENING. - -- jimoe at sohnen-moe dot com pgp/gpg public key: http://www.keyserver.net/en/ -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 5.0 OS/2 for non-commercial use Comment: PGP 5.0 for OS/2 Charset: cp850 wj8DBQE/X0RnsxxMki0foKoRArqUAKD50CzHm2NcgJyLYAPSCO9i8idiIACg4WNY cIQUOkXg96sUyPDLLPKVjNY= =c0P/ -----END PGP SIGNATURE----- _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 2 ==========================** Date: Wed, 10 Sep 2003 12:21:54 -0400 From: "T.Sikora" Subject: gcc2 Added EMXOMFLD_TYPE=link386 to gcc2_local. GCC defaults to VAC's ilink but does look for the EMXOMFLD_TYPE env setting. Replaced the missing TERMCAP and updated gmake to 381rc1-r2. -- T.Sikora tsikora at ntplx dot net _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 3 ==========================** Date: Wed, 10 Sep 2003 12:42:06 +0100 From: Stefan.Neis at t-online.de Subject: Re: Building db (sleepycat) Hi, > I don't undestand what you did. The portable solution of IFS=":" > (no matter which quotes are used) is to replace it by > IFS="${PATH_SEPARATOR-:}" Just some wild guessing: The problem might be that some shell code abuses IFS to "tokenize" some arbitrary strings, not some PATH related stuff - and then, using $PATH_SEPARATOR instead of ":" might just be wrong. Regards, Stefan _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 4 ==========================** Date: Wed, 10 Sep 2003 17:58:26 -0700 (MST) From: "James Moe" Subject: Re: Building db (sleepycat) -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Wed, 10 Sep 2003 18:12:03 +0100, Stefan.Neis at t-online.de wrote: >> >> The assignment IFS=: is IGNORED, NOT DONE, BYPASSED, BLOWN OFF, SNUBBED, >> NOT HAPPENING. > >The problem is that the assignment "IFS=:" is often used for handling >paths, so it might be automagically translated to "IFS=;" since in 95% >of the cases that is what is needed. :-( > Ah! Finally I understand. Because a colon is the default on *nix no one notices that the IFS=: does not work since that is the default. It only became a problem when I used it on os/2 where IFS=; . This broken method has shown up in two places so far. I started by building bogofilter which requires either db (sleepycat) or tdb (trivial db). db creates a config.status with the IFS=: defect. tdb generates a libtool with the defect. - -- jimoe at sohnen-moe dot com pgp/gpg public key: http://www.keyserver.net/en/ -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 5.0 OS/2 for non-commercial use Comment: PGP 5.0 for OS/2 Charset: cp850 wj8DBQE/X8iysxxMki0foKoRAl4GAJ4urMfrakOste750+YGT4CkwSGB0gCguAdj 6JaV9270qVuCi4kgeGhA0WE= =ZYZ8 -----END PGP SIGNATURE----- _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 5 ==========================** Date: Wed, 10 Sep 2003 18:12:03 +0100 From: Stefan.Neis at t-online.de Subject: Re: Building db (sleepycat) Hi, > >The problem might be that some shell code abuses IFS to "tokenize" > >some arbitrary strings, ... > > > Yes, this is exactly is happening. > However the point is this: > > The assignment IFS=: is IGNORED, NOT DONE, BYPASSED, BLOWN OFF, SNUBBED, NOT HAPPENING. The problem is that the assignment "IFS=:" is often used for handling paths, so it might be automagically translated to "IFS=;" since in 95% of the cases that is what is needed. :-( I hope I'm missing some clever trick used by Andreas, but if fear that my explanation is describing the current situation. Regards, Stefan _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 6 ==========================** Date: Wed, 10 Sep 2003 18:12:38 -0700 (MST) From: "James Moe" Subject: Possible problem with read() and write()? -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello, I finally got a package to build, tdb. There were two major EMX library problems I ran into: read() and write(). Both functions failed to operate correctly. tdb does a lot of lseek() followed by a read() or write(). read() would not always input the requested number bytes, it occasionally returned one byte short. Looking in \emx\src\lib\io\read.c there is a function that does indeed decrement the byte count by 1: read_lookahead(). I have not pursued this any further yet. write() exhibited a similar problem only it returned the correct info without actually performing correctly. By replacing the read() and write with DosRead() and DosWrite() those problem went away. So, this is a long-winded way of asking: Has any one else had problems with read() or write()? I built bogofilter + tdb using gcc 3.2.1. - -- jimoe at sohnen-moe dot com pgp/gpg public key: http://www.keyserver.net/en/ -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 5.0 OS/2 for non-commercial use Comment: PGP 5.0 for OS/2 Charset: cp850 wj8DBQE/X8wFsxxMki0foKoRAjfoAJ0VmRwf8cBq3opNbth4qcLOeOlUHACg7vmG OdYlQAPMHRufZwhYpsUKcnE= =trDi -----END PGP SIGNATURE----- _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 7 ==========================** Date: Wed, 10 Sep 2003 18:40:49 -0700 (MST) From: "James Moe" Subject: Possible problem with read() and write()? -- Never mind -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Like a bolt out of the blue, I realized that maybe it is the file's open mode that is the problem. It is. With EMX O_TEXT is the default mode. Neither bogofilter nor tdb indicate a mode, so the occasional lost byte would be a CR being chucked out. Does *nix open in binary mode by default? I guess os/2 is the same? Hmm. - -- jimoe at sohnen-moe dot com pgp/gpg public key: http://www.keyserver.net/en/ -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 5.0 OS/2 for non-commercial use Comment: PGP 5.0 for OS/2 Charset: cp850 wj8DBQE/X9KhsxxMki0foKoRAiDIAJ9fgqpHXCPn2Su54bmJzn0ow7+J6wCglyJ1 PwyHjKrwifyqRx4e6p06lUQ= =1/0Y -----END PGP SIGNATURE----- _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 8 ==========================** Date: Wed, 10 Sep 2003 23:44:56 -0400 From: Henry Sobotka Subject: Re: Possible problem with read() and write()? Do you have the mode set right (default is text with EMX read() and write(), binary with DosRead|Write)? h~ _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 9 ==========================** Date: Wed, 10 Sep 2003 23:47:06 -0400 From: Henry Sobotka Subject: Re: Possible problem with read() and write()? -- Never mind Answered my question...compiling with -Zbin-files should do the trick. h~ _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs