Re: configure is screwed!
Garst == Garst R Reese [EMAIL PROTECTED] writes: The idea I had is that you should use ac2.13+am2.15 for now, and jump to ac2.53+am1.6 when we have removed the old constructs in our configure script (in 1.3.0). Garst ac2.50 and am1.4p4 came with slackware8.0, I upgraded to 2.53 Garst and 2.15, and all worked well until now. No big deal for me to Garst back down to 2.13, but I think it is going to cause lots of Garst complaints when 1.20 is released. Garst What I wrote only applies to people like Angus (and me when I am in france) who use compaq cxx to compile. JMarc
Re: configure is screwed!
Garst == Garst R Reese [EMAIL PROTECTED] writes: Garst Jean-Marc Lasgouttes wrote: What I wrote only applies to people like Angus (and me when I am in france) who use compaq cxx to compile. Garst Well, I got the FIX FIX FIX msg. with Friday's CVS, so the Garst problem seems to have spread. Switching to 2.13 made it go Garst away, but maybe that was the make distclean etc. Garst Yes, you should not use autoconf 2.53, but only 2.13 or 2.5[012]. JMarc
Re: configure is screwed!
> "Garst" == Garst R Reese <[EMAIL PROTECTED]> writes: >> The idea I had is that you should use ac2.13+am2.15 for now, and >> jump to ac2.53+am1.6 when we have removed the old constructs in our >> configure script (in 1.3.0). Garst> ac2.50 and am1.4p4 came with slackware8.0, I upgraded to 2.53 Garst> and 2.15, and all worked well until now. No big deal for me to Garst> back down to 2.13, but I think it is going to cause lots of Garst> complaints when 1.20 is released. Garst What I wrote only applies to people like Angus (and me when I am in france) who use compaq cxx to compile. JMarc
Re: configure is screwed!
> "Garst" == Garst R Reese <[EMAIL PROTECTED]> writes: Garst> Jean-Marc Lasgouttes wrote: >> What I wrote only applies to people like Angus (and me when I am >> in france) who use compaq cxx to compile. Garst> Well, I got the FIX FIX FIX msg. with Friday's CVS, so the Garst> problem seems to have spread. Switching to 2.13 made it go Garst> away, but maybe that was the make distclean etc. Garst Yes, you should not use autoconf 2.53, but only 2.13 or 2.5[012]. JMarc
configure is screwed!
My compiler isn't happy again and this time I don't think it's being too paranoid. cxx -std strict_ansi -DHAVE_CONFIG_H -I. -I../../../devel/src/support -I../../src -I../../../devel/src/support/../ -I../../../devel -I../.. -I../../../devel/boost -I../../../devel/src/cheaders -I/usr/local/include -nocleanup -msg_display_number -msg_disable 11,193,236,261,401,611 -w0 -ptr /usr/users/aleem/OTHERS_CODE/lyx/devel-build/lyx_cxx_repository -O2 -c -MD ../../../devel/src/support/tempname.C cxx: Warning: ../../../devel/src/support/tempname.C, line 34: #117-D non-void function unnamed::make_tempfile (declared at line 19) should return a value } ^ cxx: Info: ../../../devel/src/support/tempname.C, line 19: #835-D parameter templ was never referenced int make_tempfile(char * templ) -^ I think that that warning should be an error: Index: src/support/tempname.C === RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/support/tempname.C,v retrieving revision 1.9 diff -u -p -r1.9 tempname.C --- src/support/tempname.C 21 Mar 2002 17:06:35 - 1.9 +++ src/support/tempname.C 15 Apr 2002 10:04:36 - -26,9 +26,7 int make_tempfile(char * templ) ::mktemp(templ); return ::open(templ, O_RDWR | O_CREAT | O_EXCL, S_IRUSR | S_IWUSR); #else -#ifdef WITH_WARNINGS -#warning FIX FIX FIX -#endif +#error FIX FIX FIX #endif #endif } This is what my config.h thinks: /* Define as 1 if the MKSTEMP function is declared */ #define HAVE_DECL_MKSTEMP 1 /* Define if you have the `mkstemp' function. */ /* #undef HAVE_MKSTEMP */ /* Define if you have the `mktemp' function. */ /* #undef HAVE_MKTEMP */ which is rubbish because both are defined. So, what's going on? Running configure: checking for memmove... no checking for memset... no checking for strchr... (cached) yes checking for putenv... (cached) yes checking for setenv... (cached) yes checking for mkfifo... no checking for mkstemp... no checking for mktemp... no In configure.in we have: AC_CHECK_FUNCS(memmove memset strchr putenv setenv mkfifo \ mkstemp mktemp) of these tests, mktemp, mkstemp are to be found in stdlib.h memccpy, memchr, memcmp, memcpy, memmove, memset are to be found in string.h Any ideas? Angus
Re: configure is screwed!
Angus == Angus Leeming [EMAIL PROTECTED] writes: Angus This is what my config.h thinks: /* Define as 1 if the MKSTEMP Angus function is declared */ #define HAVE_DECL_MKSTEMP 1 Angus /* Define if you have the `mkstemp' function. */ /* #undef Angus HAVE_MKSTEMP */ Angus /* Define if you have the `mktemp' function. */ /* #undef Angus HAVE_MKTEMP */ Angus which is rubbish because both are defined. So, what's going on? Do you run with a version of autoconf which cannot find functions when in C++ language (ac 2.5[012], I think) with compaq cxx? Otherwise, what are the error messages in config.log? JMarc
Re: configure is screwed!
On Monday 15 April 2002 11:21 am, Jean-Marc Lasgouttes wrote: Angus == Angus Leeming [EMAIL PROTECTED] writes: Angus This is what my config.h thinks: /* Define as 1 if the MKSTEMP Angus function is declared */ #define HAVE_DECL_MKSTEMP 1 Angus /* Define if you have the `mkstemp' function. */ /* #undef Angus HAVE_MKSTEMP */ Angus /* Define if you have the `mktemp' function. */ /* #undef Angus HAVE_MKTEMP */ Angus which is rubbish because both are defined. So, what's going on? Do you run with a version of autoconf which cannot find functions when in C++ language (ac 2.5[012], I think) with compaq cxx? Otherwise, what are the error messages in config.log? JMarc Sign... Yes, I'm running with ac 2.52. Should I downgrade to 2.13? Angus configure:17472: checking for mktemp configure:17509: cxx -std strict_ansi -o conftest -nocleanup -msg_display_number -msg_disable 11,193,236,261,401,611 -w0 -ptr /usr/users/aleem/OTHERS_CODE/lyx/devel-build/lyx_cxx_repository -O2 -I$(top_srcdir)/src/cheaders -I/usr/local/include-L/usr/local/lib conftest.cc -ljpeg -lSM -lICE -lc -lm -lX11 -lbsd 5 cxx: Error: configure, line 17488: #338 more than one instance of overloaded function mktemp has C linkage char mktemp (); -^ cxx: Error: configure, line 17500: #386 no instance of overloaded function mktemp matches the required type f = mktemp; --^ cxx: Info: 2 errors detected in the compilation of conftest.cc. configure:17512: $? = 1 configure: failed program was: #line 17478 configure #include confdefs.h /* System header to define __stub macros and hopefully few prototypes, which can conflict with char mktemp (); below. */ #include assert.h /* Override any gcc2 internal prototype to avoid an error. */ #ifdef __cplusplus extern C #endif /* We use char because int might match the return type of a gcc2 builtin and then its argument prototype would still apply. */ char mktemp (); char (*f) (); int main () { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub_mktemp) || defined (__stub___mktemp) choke me #else f = mktemp; #endif ; return 0; } configure:17528: result: no
Re: configure is screwed!
Angus == Angus Leeming [EMAIL PROTECTED] writes: Angus Sign... Yes, I'm running with ac 2.52. Should I downgrade to Angus 2.13? Angus This is the reason why I was not in favor of using AC_LANG_C for xforms stuff, since every function detection will fail. And switching to C for all function detection is not good, since we cannot guarantee that some of the other ,4 files will never do that. The idea I had is that you should use ac2.13+am2.15 for now, and jump to ac2.53+am1.6 when we have removed the old constructs in our configure script (in 1.3.0). JMarc
Re: configure is screwed!
On Monday 15 April 2002 4:04 pm, Jean-Marc Lasgouttes wrote: Angus == Angus Leeming [EMAIL PROTECTED] writes: Angus Sign... Yes, I'm running with ac 2.52. Should I downgrade to Angus 2.13? Angus This is the reason why I was not in favor of using AC_LANG_C for xforms stuff, since every function detection will fail. And switching to C for all function detection is not good, since we cannot guarantee that some of the other ,4 files will never do that. The idea I had is that you should use ac2.13+am2.15 for now, I'm now using ac 2.13 and am 1.5 and all is fine. Angus and jump to ac2.53+am1.6 when we have removed the old constructs in our configure script (in 1.3.0). JMarc
Re: configure is screwed!
Garst R. Reese [EMAIL PROTECTED] writes: The idea I had is that you should use ac2.13+am2.15 for now, and jump to ac2.53+am1.6 when we have removed the old constructs in our configure script (in 1.3.0). JMarc | ac2.50 and am1.4p4 came with slackware8.0, I upgraded to 2.53 and 2.15, | and all worked well until now. | No big deal for me to back down to 2.13, but I think it is going to | cause lots of complaints when 1.20 is released. Why? users compiling a tar.gz dist will just use the provided configure. -- Lgb
configure is screwed!
My compiler isn't happy again and this time I don't think it's being too paranoid. cxx -std strict_ansi -DHAVE_CONFIG_H -I. -I../../../devel/src/support -I../../src -I../../../devel/src/support/../ -I../../../devel -I../.. -I../../../devel/boost -I../../../devel/src/cheaders -I/usr/local/include -nocleanup -msg_display_number -msg_disable 11,193,236,261,401,611 -w0 -ptr /usr/users/aleem/OTHERS_CODE/lyx/devel-build/lyx_cxx_repository -O2 -c -MD ../../../devel/src/support/tempname.C cxx: Warning: ../../../devel/src/support/tempname.C, line 34: #117-D non-void function "::make_tempfile" (declared at line 19) should return a value } ^ cxx: Info: ../../../devel/src/support/tempname.C, line 19: #835-D parameter "templ" was never referenced int make_tempfile(char * templ) -^ I think that that warning should be an error: Index: src/support/tempname.C === RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/support/tempname.C,v retrieving revision 1.9 diff -u -p -r1.9 tempname.C --- src/support/tempname.C 21 Mar 2002 17:06:35 - 1.9 +++ src/support/tempname.C 15 Apr 2002 10:04:36 - @@ -26,9 +26,7 @@ int make_tempfile(char * templ) ::mktemp(templ); return ::open(templ, O_RDWR | O_CREAT | O_EXCL, S_IRUSR | S_IWUSR); #else -#ifdef WITH_WARNINGS -#warning FIX FIX FIX -#endif +#error FIX FIX FIX #endif #endif } This is what my config.h thinks: /* Define as 1 if the MKSTEMP function is declared */ #define HAVE_DECL_MKSTEMP 1 /* Define if you have the `mkstemp' function. */ /* #undef HAVE_MKSTEMP */ /* Define if you have the `mktemp' function. */ /* #undef HAVE_MKTEMP */ which is rubbish because both are defined. So, what's going on? Running configure: checking for memmove... no checking for memset... no checking for strchr... (cached) yes checking for putenv... (cached) yes checking for setenv... (cached) yes checking for mkfifo... no checking for mkstemp... no checking for mktemp... no In configure.in we have: AC_CHECK_FUNCS(memmove memset strchr putenv setenv mkfifo \ mkstemp mktemp) of these tests, mktemp, mkstemp are to be found in stdlib.h memccpy, memchr, memcmp, memcpy, memmove, memset are to be found in string.h Any ideas? Angus
Re: configure is screwed!
> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes: Angus> This is what my config.h thinks: /* Define as 1 if the MKSTEMP Angus> function is declared */ #define HAVE_DECL_MKSTEMP 1 Angus> /* Define if you have the `mkstemp' function. */ /* #undef Angus> HAVE_MKSTEMP */ Angus> /* Define if you have the `mktemp' function. */ /* #undef Angus> HAVE_MKTEMP */ Angus> which is rubbish because both are defined. So, what's going on? Do you run with a version of autoconf which cannot find functions when in C++ language (ac 2.5[012], I think) with compaq cxx? Otherwise, what are the error messages in config.log? JMarc
Re: configure is screwed!
On Monday 15 April 2002 11:21 am, Jean-Marc Lasgouttes wrote: > > "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes: > > Angus> This is what my config.h thinks: /* Define as 1 if the MKSTEMP > Angus> function is declared */ #define HAVE_DECL_MKSTEMP 1 > > Angus> /* Define if you have the `mkstemp' function. */ /* #undef > Angus> HAVE_MKSTEMP */ > > Angus> /* Define if you have the `mktemp' function. */ /* #undef > Angus> HAVE_MKTEMP */ > > Angus> which is rubbish because both are defined. So, what's going on? > > Do you run with a version of autoconf which cannot find functions when > in C++ language (ac 2.5[012], I think) with compaq cxx? > > Otherwise, what are the error messages in config.log? > > JMarc Sign... Yes, I'm running with ac 2.52. Should I downgrade to 2.13? Angus configure:17472: checking for mktemp configure:17509: cxx -std strict_ansi -o conftest -nocleanup -msg_display_number -msg_disable 11,193,236,261,401,611 -w0 -ptr /usr/users/aleem/OTHERS_CODE/lyx/devel-build/lyx_cxx_repository -O2 -I$(top_srcdir)/src/cheaders -I/usr/local/include-L/usr/local/lib conftest.cc -ljpeg -lSM -lICE -lc -lm -lX11 -lbsd >&5 cxx: Error: configure, line 17488: #338 more than one instance of overloaded function "mktemp" has "C" linkage char mktemp (); -^ cxx: Error: configure, line 17500: #386 no instance of overloaded function "mktemp" matches the required type f = mktemp; --^ cxx: Info: 2 errors detected in the compilation of "conftest.cc". configure:17512: $? = 1 configure: failed program was: #line 17478 "configure" #include "confdefs.h" /* System header to define __stub macros and hopefully few prototypes, which can conflict with char mktemp (); below. */ #include /* Override any gcc2 internal prototype to avoid an error. */ #ifdef __cplusplus extern "C" #endif /* We use char because int might match the return type of a gcc2 builtin and then its argument prototype would still apply. */ char mktemp (); char (*f) (); int main () { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub_mktemp) || defined (__stub___mktemp) choke me #else f = mktemp; #endif ; return 0; } configure:17528: result: no
Re: configure is screwed!
> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes: Angus> Sign... Yes, I'm running with ac 2.52. Should I downgrade to Angus> 2.13? Angus This is the reason why I was not in favor of using AC_LANG_C for xforms stuff, since every function detection will fail. And switching to C for all function detection is not good, since we cannot guarantee that some of the other ,4 files will never do that. The idea I had is that you should use ac2.13+am2.15 for now, and jump to ac2.53+am1.6 when we have removed the old constructs in our configure script (in 1.3.0). JMarc
Re: configure is screwed!
On Monday 15 April 2002 4:04 pm, Jean-Marc Lasgouttes wrote: > > "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes: > > Angus> Sign... Yes, I'm running with ac 2.52. Should I downgrade to > Angus> 2.13? Angus > > This is the reason why I was not in favor of using AC_LANG_C for > xforms stuff, since every function detection will fail. And switching > to C for all function detection is not good, since we cannot > guarantee that some of the other ,4 files will never do that. > > The idea I had is that you should use ac2.13+am2.15 for now, I'm now using ac 2.13 and am 1.5 and all is fine. Angus > and jump to ac2.53+am1.6 when we have removed the old constructs in our > configure script (in 1.3.0). > > JMarc
Re: configure is screwed!
"Garst R. Reese" <[EMAIL PROTECTED]> writes: >> The idea I had is that you should use ac2.13+am2.15 for now, and jump >> to ac2.53+am1.6 when we have removed the old constructs in our >> configure script (in 1.3.0). >> >> JMarc | ac2.50 and am1.4p4 came with slackware8.0, I upgraded to 2.53 and 2.15, | and all worked well until now. | No big deal for me to back down to 2.13, but I think it is going to | cause lots of complaints when 1.20 is released. Why? users compiling a tar.gz dist will just use the provided configure. -- Lgb