Date: Fri, 4 Apr 2003 02:40:34 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [Ux2bs_Archive] No. 124 ************************************************** Thursday 03 April 2003 Number 124 ************************************************** Subjects for today 1 Re: Building GMP (GNU MP) : sma at sohnen-moe.com 2 Suggested changes to build.sh : sma at sohnen-moe.com 3 Re: Building GMP (GNU MP) : sma at sohnen-moe.com 4 Re: Building GMP (GNU MP) : sma at sohnen-moe.com 5 Re: Procmail : sma at sohnen-moe.com 6 Procmail : John Poltorak 7 Re: Building GMP (GNU MP) : Sebastian Wittmeier (ShadoW)" **= Email 1 ==========================** Date: Fri, 04 Apr 2003 13:00:11 -0700 From: sma at sohnen-moe.com Subject: Re: Building GMP (GNU MP) This is a MIME encapsulated message. --===_3E8DEFA0_== Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit >Anyone tried building GMP? > Yes, I got it to build, v4.1.2, with both gcc v2.8.1 and v3.2.1. It does not include posix2. There are a small number of problems: 1. autoconf spits out 3 errors about a missing path. These appear irrelevant. Using:- autoconf (GNU Autoconf) 2.57 (release for OS/2) SYS0003: The system cannot find the path specified. SYS0003: The system cannot find the path specified. SYS0003: The system cannot find the path specified. configure.in:58: h:/usr/bin/m4: Bad expression in eval: > 0 AUTOM4TE.: h:/usr/bin/m4.exe failed with exit status: 1 ./configure 2. configure must be changed. It uses "ln" instead of "$as_ln_s" in two places. A diff file is attached. 3. Near the end of the build, while creating the libraries, something crashes with a SIGSEGV (sh.exe? ld.exe? Nothing in popupos2.log). By cd'ing to the build directory and entering "make" the build finishes without a problem. /bin/sh ./libtool --mode=link gcc -o libgmp.la -rpath /usr/local/lib -version-info 6:2:3 assert.lo compat.lo errno.lo extract-dbl.lo memory.lo mp_bpl.lo mp_clz_tab.lo mp_minv_tab.lo [... a whoooole lot more ...] scanf/vfscanf.lo scanf/vscanf.lo scanf/vsscanf.lo Process terminated by SIGSEGV make[2]: *** [libgmp.la] Error 3 The command line is huge: 5095 characters long. However, this seems not to matter when cd'ing to the build dir and running make directly. "make check" passed its tests except for those in test/misc: a compilation problem with localeconv. "make install" proceeds correctly. --===_3E8DEFA0_== Content-Type: application/octet-stream; name="gmp-4.1.2-configure.diff" Content-Transfer-Encoding: base64 LS0tIGNvbmZpZ3VyZS1vcmlnCTIwMDItMTItMjcgMTY6MzA6MjYuMDAwMDAwMDAwIC0wNzAw DQorKysgY29uZmlndXJlLWxuLWZpeAkyMDAzLTA0LTA0IDEyOjUyOjA4LjAwMDAwMDAwMCAt MDcwMA0KQEAgLTIyNjE4LDggKzIyNjE4LDggQEANCiAgIGVzYWMNCiANCiAgICMgTWFrZSBh IHN5bWxpbmsgaWYgcG9zc2libGU7IG90aGVyd2lzZSB0cnkgYSBoYXJkIGxpbmsuDQotICBs biAtcyAkYWNfcmVsX3NvdXJjZSAkYWNfZGVzdCAyPi9kZXYvbnVsbCB8fA0KLSAgICBsbiAk c3JjZGlyLyRhY19zb3VyY2UgJGFjX2Rlc3QgfHwNCisgICRhc19sbl9zICRhY19yZWxfc291 cmNlICRhY19kZXN0IDI+L2Rldi9udWxsIHx8DQorICAgICRhc19sbl9zICRzcmNkaXIvJGFj X3NvdXJjZSAkYWNfZGVzdCB8fA0KICAgICB7IHsgZWNobyAiJGFzX21lOjIyNjIzOiBlcnJv cjogY2Fubm90IGxpbmsgJGFjX2Rlc3QgdG8gJHNyY2Rpci8kYWNfc291cmNlIiA+JjUNCiBl Y2hvICIkYXNfbWU6IGVycm9yOiBjYW5ub3QgbGluayAkYWNfZGVzdCB0byAkc3JjZGlyLyRh Y19zb3VyY2UiID4mMjt9DQogICAgeyAoZXhpdCAxKTsgZXhpdCAxOyB9OyB9DQo= --===_3E8DEFA0_== Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit application/octet-stream base64>- --===_3E8DEFA0_== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs --===_3E8DEFA0_==-- **= Email 2 ==========================** Date: Fri, 04 Apr 2003 13:14:52 -0700 From: sma at sohnen-moe.com Subject: Suggested changes to build.sh This is a MIME encapsulated message. --===_3E8DE884_== Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Hello, Attached is a modified build.sh. The added changes are either error checking or more informational output. The error checks stop the build if something does not go well, like configure failing, or build.table has bogus info, or make stops unexpectedly, etc. --===_3E8DE884_== Content-Type: application/octet-stream; name="build.sh" Content-Transfer-Encoding: base64 UEtHPSQxDQpleHBvcnQgT1MyUEFUSD0kUEFUSA0KZXhwb3J0IFBBVEg9YGVjaG8gLUUgJFBB VEggfHRyICdcXFxcJyAnLydgDQpkYXRlDQpzdGFydF90aW1lPWBkYXRlICsiJXMiYA0KZWNo byAkU0hFTEwNCg0KIw0KIyAgUGVybCBkb2VzIG5vdCBsaWtlIHRoaXMgbGluZTotDQojDQoj ZXhwb3J0IEVNWFNIRUxMPSR1eHJ0L2Jpbi9zaA0KDQojDQojICBEZWZpbmUgc3VuZHJ5IGZ1 bmN0aW9ucw0KIw0KDQpmaW5pc2goKQ0Kew0KICBlY2hvDQojICBlbnYNCiAgZWNobyAiLi5D b21waWxlciB2ZXJzaW9uIGlzOiBgZ2NjIC0tdmVyc2lvbiB8IGhlYWQgLW4xYCI7DQogIGRh dGUNCiAgZW5kX3RpbWU9YGRhdGUgKyIlcyJgDQogIGVjaG8gZWxhcHNlZCB0aW1lOiBgZXhw ciAkZW5kX3RpbWUgLSAkc3RhcnRfdGltZWAgc2Vjcw0KICBlY2hvIGVuZA0KfQ0KDQppZiB0 ZXN0ICEgLWYgJE9TUlQvb3MyL2xpbmszODYuZXhlIDsgdGhlbg0Kew0KICBlY2hvICJUaGVy ZSBpcyBubyBMSU5LMzg2LkVYRSBvbiB0aGUgcGF0aCwgd2hpY2ggY291bGQgY2F1c2UgcHJv YmxlbXMuIg0KICBlY2hvICJQcmVzcyBFTlRFUiB0byBjb250aW51ZSINCiAgcmVhZCByZXBs eQ0KfQ0KZmkNCg0KZ3JlcCBeJFBLR1w7IGJ1aWxkLnRhYmxlID4gJFRNUC91eDJfdG1wDQoN CklGUz0nOycNCg0KaWYgWyAtcyAgJFRNUC91eDJfdG1wIF07IHRoZW4NCiAgZWNobyAiJFBL RyBidWlsZCBpbmZvOiBbYGNhdCAkVE1QL3V4Ml90bXBgXSI7DQogIHJlYWQgcDEgRElSIFVS TCBDRkxBR1MgTERGTEFHUyBDRkdQQVJNUyBNQUtFUEFSTSBTUkMgPCAkVE1QL3V4Ml90bXA7 DQplbHNlDQogIGVjaG8gIiAkVE1QL3V4Ml90bXAgZG9lcyBub3QgZXhpc3Qgb3IgaXMgZW1w dHkuIFRoaXMgaXMgYmFkIHRoaW5nLiI7DQogIGZpbmlzaDsNCiAgZXhpdCAxOw0KZmkNCg0K DQplY2hvIERJUiBVUkwgQ0ZMQUdTIExERkxBR1MgQ0ZHUEFSTVMgTUFLRVBBUk0gU1JDDQpl Y2hvICRESVIgJFVSTCAkQ0ZMQUdTICRMREZMQUdTICRDRkdQQVJNUyAkTUFLRVBBUk0gJFNS Qw0KDQplY2hvIFVSTCAkVVJMDQplY2hvIENGTEFHUyAkQ0ZMQUdTDQplY2hvIExERkxBR1Mg JExERkxBR1MNCmVjaG8gQ0ZHUEFSTVMgJENGR1BBUk1TDQplY2hvIE1BS0VQQVJNICRNQUtF UEFSTQ0KZWNobyBTUkMgJFNSQw0KDQpleHBvcnQgQ0ZMQUdTDQpleHBvcnQgTERGTEFHUw0K DQoNCmlmIFsgLXogJFVSTCBdOyB0aGVuDQogIGVjaG8gIlVSTCBpcyBub3QgZGVmaW5lZC4g SXQgaXMgcmVxdWlyZWQuIg0KICBmaW5pc2g7DQogIGV4aXQgMTsNCmZpDQoNCkFSQ0hJVkVG SUxFPWBiYXNlbmFtZSAkVVJMYA0KQVJDSElWRT1gYmFzZW5hbWUgJFVSTCB8IHNlZCAncy9c LnQuKnovLydgDQoNCnRlc3QgJERJUiB8fCBESVI9JEFSQ0hJVkUNCg0KZWNobyBSRVBPU0lU T1JZL0FSQ0hJVkVGSUxFID0gJFJFUE9TSVRPUlkvJEFSQ0hJVkVGSUxFDQppZiB0ZXN0IC1m ICRSRVBPU0lUT1JZLyRBUkNISVZFRklMRTsgdGhlbg0KICBlY2hvICIuLiRBUkNISVZFRklM RSBhbHJlYWR5IHJldHJpZXZlZC4iOw0KZWxzZQ0KICB3Y21kPSJ3Z2V0IC1udiAtLW5vLWNs b2JiZXIgLS10cmllcz0xIC1QICRSRVBPU0lUT1JZICRVUkwiDQogIGVjaG8gIi4ucmV0cmll dmluZyAkQVJDSElWRUZJTEUgKCR3Y21kKSINCiAgZXZhbCAkd2NtZA0KZmkNCg0KaWYgWyAh IC1mICRSRVBPU0lUT1JZLyRBUkNISVZFRklMRSBdOyB0aGVuDQogIGVjaG8gIi4ud2dldCBv ZiAkUkVQT1NJVE9SWS8kQVJDSElWRUZJTEUgZmFpbGVkIjsNCiAgZmluaXNoOw0KICBleGl0 IDE7DQpmaQ0KDQoNCklGUz0nICcNCg0KY2QgJFdPUktESVINCnB3ZA0KdGVzdCAtZCAkRElS IHx8IHRhciB6eGYgJFJFUE9TSVRPUlkvJEFSQ0hJVkVGSUxFDQpjZCAkRElSDQpjaG1vZCAt UiBhK3cgKg0KcHdkDQoNCiMNCiMgQ2hlY2sgZm9yIGZpeC11cHMsIHBhdGNoZXMuDQojDQpQ QVRDSEZJTEU9JEJMRF9IT01FL3BhdGNoZXMvJEFSQ0hJVkUuZGlmZg0KaWYgdGVzdCAtZiAk UEFUQ0hGSUxFOyB0aGVuDQp7DQogIHBhdGNoIC1wYGF3ayAnQkVHSU4gew0KICAgICAgbWlu X3AgPSAxMjM0NTY3DQogICAgfQ0KICAgICQxID09ICIrKysiIHx8ICQxID09ICIqKioiIHsN CiAgICAgIGdzdWIgKEFSQ0hJVkUgIi4qIiwgIiIsICQyKQ0KICAgICAgZ3N1YiAoL1teL10v LCAiIiwgJDIpDQogICAgICBpZiAobGVuZ3RoKCQyKSA8IG1pbl9wKSBtaW5fcCA9IGxlbmd0 aCgkMikNCiAgICB9DQoNCiAgRU5EIHsNCiAgICBwcmludCBtaW5fcCArIDENCiAgfScgQVJD SElWRT0kQVJDSElWRSAkUEFUQ0hGSUxFYCA8ICRQQVRDSEZJTEUNCn0NCmZpDQoNCiMNCiMg UHJlLXByb2Nlc3NpbmcuDQojDQpmbmFtZT0kQkxEX0hPTUUvc2NyaXB0cy9wcmUtcHJvY2Vz cy8kUEtHDQppZiBbIC1mICRmbmFtZSBdOyB0aGVuDQogIGVjaG8gIi4uVXNpbmcgYWRkaXRp b25hbCBzaXRlIGFzc2lnbm1lbnRzICgkZm5hbWUpIg0KICAuICRmbmFtZQ0KZmkNCg0KZm5h bWU9JEJMRF9IT01FL3NjcmlwdHMvYnVpbGQvYnVpbGRfJDEuY21kDQppZiB0ZXN0IC1mICRm bmFtZTsgdGhlbg0KICB7DQogIGVjaG8gIi4uVXNpbmcgJDEgc3BlY2lmaWMgQnVpbGQgc2Ny aXB0ICgkZm5hbWUpIg0KICBleHBvcnQgQ09ORklHX1NJVEU9DQogIGV4cG9ydCBQQVRIPSRP UzJQQVRIDQogICRCTERfSE9NRS9zY3JpcHRzL2J1aWxkL2J1aWxkXyQxLmNtZA0KICBmaW5p c2gNCiAgZXhpdCAxDQogIH0NCmZpDQoNCiMNCiMgVXNlIGEgY3VzdG9tIG9zLzIgbWFrZWZp bGU/DQojDQpmbmFtZT0kQkxEX0hPTUUvbWFrZWZpbGVzLyRBUkNISVZFLm1haw0KaWYgdGVz dCAtZiAkZm5hbWU7IHRoZW4NCiAgew0KICBlY2hvICJ1c2luZyAkMSBzcGVjaWZpYyBNYWtl ZmlsZSAoJGZuYW1lKSINCiAgbWFrZSAtZiAkQkxEX0hPTUUvbWFrZWZpbGVzLyRBUkNISVZF Lm1haw0KICBtYWtlIC1mICRCTERfSE9NRS9tYWtlZmlsZXMvJEFSQ0hJVkUubWFrIGluc3Rh bGwNCiAgZmluaXNoDQogIGV4aXQgMQ0KICB9DQoNCmZpDQoNCmZuYW1lPSRCTERfSE9NRS9w YXRjaGVzLyRBUkNISVZFLU9TMi1wYXRjaC56aXANCmlmIHRlc3QgLWYgJGZuYW1lIDsgdGhl bg0KICB7DQogIGVjaG8gIi4uVXNpbmcgU0FXQVRBSVNISSBwYXRjaCBraXQgKCRmbmFtZSki DQogIHVuemlwICRCTERfSE9NRS9wYXRjaGVzLyRBUkNISVZFLU9TMi1wYXRjaC56aXAgT1My cGF0Y2guemlwDQogIHVuemlwIE9TMnBhdGNoLnppcCAmJiBybSBPUzJwYXRjaC56aXANCiAg UEFUSD0kUlQvJzsnJFBBVEgNCiAgZXhwb3J0IFBBVEgNCiAgZXhwb3J0IENPTkZJR19TSVRF PW9zMi9jb25maWcuc2l0ZQ0KICBvczJ1bml4IC1BTEwNCiAgfQ0KZmkNCg0KZm5hbWU9JEJM RF9IT01FL3NjcmlwdHMvcHJlLWNvbmYvJFBLRw0KaWYgdGVzdCAtZiAkZm5hbWU7IHRoZW4N CiAgZWNobyAiLi4gVXNpbmcgcHJlLWNvbmYgKCRmbmFtZSkiOw0KICAuICRmbmFtZQ0KZWxz ZQ0KCWVjaG8gVXNpbmc6LSBgYXV0b2NvbmYgLS12ZXJzaW9uIHwgaGVhZCAtMWANCglhdXRv Y29uZg0KZmkNCg0KDQplY2hvIC4vY29uZmlndXJlICRDRkdQQVJNUw0KLi9jb25maWd1cmUg JENGR1BBUk1TDQoNCmlmIFsgMCAhPSAkPyBdOyB0aGVuDQogIGVjaG8gIi4uY29uZmlndXJl IGZhaWxlZC4iOw0KICBmaW5pc2g7DQogIGV4aXQgMTsNCmZpDQoNCmVjaG8gY3AgLXAgJEJM RF9IT01FL2xpYi9hdXgvbWtpbnN0YWxsZGlycw0KY3AgLXAgJEJMRF9IT01FL2xpYi9hdXgv bWtpbnN0YWxsZGlycyAuDQoNCmVjaG8gbWFrZSAkTUFLRVBBUk0NCmVjaG8gLi4NCmlmIHRl c3QgJE1BS0VQQVJNOyB0aGVuDQogIG1ha2UgIiRNQUtFUEFSTSINCmVsc2UNCiAgbWFrZQ0K ZmkNCnJjPSQ/DQplY2hvICIuLm1ha2UgcmV0dXJuIGlzICRyYyI7DQoNCmlmIFsgMCAhPSAk cmMgXTsgdGhlbg0KICBlY2hvICIuLm1ha2UgZmFpbGVkLiI7DQogIGZpbmlzaDsNCiAgZXhp dCAxOw0KZmkNCg0KbWFrZSBpbnN0YWxsDQoNCnRlc3QgLWYgJEJMRF9IT01FL3NjcmlwdHMv cG9zdC1wcm9jZXNzLyRQS0cgJiYgLiAkQkxEX0hPTUUvc2NyaXB0cy9wb3N0LXByb2Nlc3Mv JFBLRw0KDQpmaW5pc2gNCg0K --===_3E8DE884_== Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit base64>- --===_3E8DE884_== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs --===_3E8DE884_==-- **= Email 3 ==========================** Date: Fri, 04 Apr 2003 14:43:47 -0700 From: sma at sohnen-moe.com Subject: Re: Building GMP (GNU MP) >> The command line is huge: 5095 characters long. However, this seems >>not to matter when cd'ing to the build dir and running make directly. > >Sometimes (probably not in this case) a second make call succeeds in >spite of errors in the first run, because a (corrupt) output file has >begun to be built. > The test runs seem quite happy with the results so the libs are not corrupted. I am looking into the compilation error I have. It looks like an imcompatibilty in the EMX environment. In /emx/include/locale.h there is: struct lconv * _localeconv (int); struct lconv * localeconv (void); #if defined (__CHAR_UNSIGNED__) #define localeconv() _localeconv(255) #else #define localeconv() _localeconv(127) #endif which is fundamentally at odds with the code in test/misc/t-locale.c /* Replace the libc localeconv with one we can manipulate. */ struct lconv * localeconv() { static struct lconv l; l.decimal_point = decimal_point; return &l; } 1. Leaving it #define'd causes an error since localeconv() expands to _localeconv(255), an error since "255" is not a valid type. 2. #undef'ing localeconv() causes a "multiply defined" error from the linker. 3. Adding a #ifndef __EMX__ conditional works: #ifndef __EMX__ /* Replace the libc localeconv with one we can manipulate. */ struct lconv * localeconv() { static struct lconv l; l.decimal_point = decimal_point; return &l; } #endif but this is an ugly hack and produces this when the tests are executed: PASS: t-printf.exe PASS: t-scanf.exe Test skipped, replacing localeconv doesn't work PASS: t-locale.exe _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 4 ==========================** Date: Fri, 04 Apr 2003 15:07:14 -0700 From: sma at sohnen-moe.com Subject: Re: Building GMP (GNU MP) > >gcc called by libtool would appear in the log file, wouldn't it? So the >crash is in libtool. > Since libtool is a shell script, I guess the crash is in sh.exe. _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 5 ==========================** Date: Fri, 04 Apr 2003 15:31:05 -0700 From: sma at sohnen-moe.com Subject: Re: Procmail > >Is anyone able to build procmail? > Have you read the FAQ? There was a question about porting to NT. There were 3 main issues that made an NT port very difficult but two of the issues are irrelevant for os/2 1. user permissions: os/2 is a single user machine, no problem. 2. fork()/exec() exists in the EMX environment. The third (set of) problem may be a showstpper. As Bart Schaefer said: "Finally, procmail relies on a host of other minor UNIX system facilities, such as hard linked files and other file permission details, which can be worked around but lack of which has either inefficient or insecure side- effects. ..." Since only hard links is actually mentioned (permissions again are irrelevant), it is difficult to say whether this matters. _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 6 ==========================** Date: Fri, 4 Apr 2003 16:26:59 +0000 From: John Poltorak Subject: Procmail Is anyone able to build procmail? You would need to add this entry to build.table to try it:- procmail;;ftp://ftp.procmail.net/pub/procmail/procmail-3.22.tar.gz;;;;;. -- John _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs **= Email 7 ==========================** Date: Fri, 04 Apr 2003 23:13:26 +0200 (CEST) From: "Sebastian Wittmeier (ShadoW)" Subject: Re: Building GMP (GNU MP) On Fri, 04 Apr 2003 13:00:11 -0700, sma at sohnen-moe.com wrote: > /bin/sh ./libtool --mode=link gcc -o libgmp.la -rpath /usr/local/lib > -version-info 6:2:3 assert.lo compat.lo errno.lo extract-dbl.lo > memory.lo mp_bpl.lo mp_clz_tab.lo mp_minv_tab.lo > [... a whoooole lot more ...] > scanf/vfscanf.lo scanf/vscanf.lo scanf/vsscanf.lo > Process terminated by SIGSEGV > make[2]: *** [libgmp.la] Error 3 gcc called by libtool would appear in the log file, wouldn't it? So the crash is in libtool. > The command line is huge: 5095 characters long. However, this seems >not to matter when cd'ing to the build dir and running make directly. Sometimes (probably not in this case) a second make call succeeds in spite of errors in the first run, because a (corrupt) output file has begun to be built. You are reworking build.sh. I have a suggestion: To ease bug tracking, could there be an option to stop build.sh at some point, or execute only individual build steps? Subsequent steps could mess up the working directory. Sebastian _______________________________________________ UX2BS mailing list UX2BS at os2ports.com http://os2ports.com/mailman/listinfo/ux2bs