Re: Problems compiling apr on Linux
On 12 Dec 2000, Jeff Trawick wrote: Mo DeJong [EMAIL PROTECTED] writes: Hi all. I tried to compile apr under Linux (Red Hat 6.2) today, but it failed with the following error: make[1]: Entering directory `/usr/local/project/subversion/apr/lib' gcc -DLINUX=2 -pthread -D_REENTRANT -c -I../include -I../include/arch/unix -I../include/arch/unix apr_pools.c In file included from ../include/apr_portable.h:70, from apr_pools.c:65: ../include/apr_network_io.h:127: parse error ../include/apr_network_io.h:175: parse error make[1]: *** [apr_pools.o] Error 1 make[1]: Leaving directory `/usr/local/project/subversion/apr/lib' What is APR_HAVE_IPV6 set to in your copy of the generated file apr.h? It is supposed to always be set to 0 or 1, hence the #if APR_HAVE_IPV6 syntax instead of #ifdef APR_HAVE_IPV6 syntax. On my RedHat 6.0 system it is set to 1 in apr.h. I am guessing that it is set to some bogus value on your system (like @have_ipv6@ or similar, indicating that something bad happened during configure). Yup, that looks like the problem. #define APR_HAVE_MEMMOVE1 #define APR_HAVE_BZERO 1 #define APR_HAVE_IPV6 @have_ipv6@ I reran autoconf, did a ./config.status --recheck ; ./config.status and now the problem seems to have fixed itself. Of course, now I am seeing another error: gcc -DLINUX=2 -pthread -D_REENTRANT -c -I../include -I../include/arch/unix apr_md5.c apr_md5.c: In function `apr_validate_password': apr_md5.c:675: warning: assignment makes pointer from integer without a cast gcc -DLINUX=2 -pthread -D_REENTRANT -c -I../include -I../include/arch/unix apr_getpass.c apr_getpass.c: In function `getpass': apr_getpass.c:107: `stderr' undeclared (first use in this function) apr_getpass.c:107: (Each undeclared identifier is reported only once apr_getpass.c:107: for each function it appears in.) make[1]: *** [apr_getpass.o] Error 1 I am not sure if this patch is correct, but it seems to fix this compile problem for me. Index: apr_getpass.c === RCS file: /home/cvspublic/apr/passwd/apr_getpass.c,v retrieving revision 1.4 diff -u -r1.4 apr_getpass.c --- apr_getpass.c 2000/12/02 12:06:03 1.4 +++ apr_getpass.c 2000/12/13 04:09:53 @@ -74,6 +74,10 @@ #include stdlib.h #endif +#if APR_HAVE_STDIO_H +#include stdio.h /* for stderr */ +#endif + #if defined(HAVE_TERMIOS_H) !defined(HAVE_GETPASS) #include termios.h #endif After that, I am seeing this new error: gcc -DLINUX=2 -pthread -D_REENTRANT -c -I../../include -I../../include/arch/unix -I../../include/arch/unix dir.c dir.c: In function `apr_dir_entry_size': dir.c:168: storage size of `filestat' isn't known Mo DeJong Red Hat Inc
Re: Problems compiling apr on Linux
Mo DeJong wrote: I reran autoconf, did a ./config.status --recheck ; ./config.status and now the problem seems to have fixed itself. You shouldn't be running autodonf directly. Instead, you should run the buildconf script in the top-level dir. Just try that, without all the patches; I think your problems will just magically go away then. -- Brane ibej home: [EMAIL PROTECTED] http://www.xbc.nu/brane/ work: [EMAIL PROTECTED] http://www.hermes-softlab.com/ ACM: [EMAIL PROTECTED]http://www.acm.org/
Re: cvs commit: apr/shmem/unix/mm aclocal.m4
[EMAIL PROTECTED] writes: dreid 00/12/13 03:09:49 Modified:shmem/unix/mm aclocal.m4 Log: We need to define this correctly or it's never set and we don't set the correct values. This caused the tests to fail on BeOS and maybe other systems as well. Sent to rse, right? -- Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site: http://www.geocities.com/SiliconValley/Park/9289/ Born in Roswell... married an alien...
Re: Showstoppers: Alpha 9
I would like to see more opinions on the src/ thing than myself, OtherBill, and Ryan. We should not proceed on *any* course of action without that. I suspect we will not have it resolved by tomorrow. I'm going to work on a bunch of the items for the release, but we should do something to get a write-up of the src-vs-not alternatives and get some more input. I think it is goofy to place a src directory in a source tree -- everything in the distribution is source. If there are too many subdirectories, then either abstract them into relevant categories or split them into different library modules. Roy
Re: cvs commit: apr/shmem/unix/mm aclocal.m4
David Reid [EMAIL PROTECTED] writes: Yeah, and he denied all knowledge of the piece of code affected :) Hence I submitted the fix! c00l... I guess that was in the temporary patch from Ryan... -- Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site: http://www.geocities.com/SiliconValley/Park/9289/ Born in Roswell... married an alien...
Re: Problems compiling apr on Linux
On Wed, Dec 13, 2000 at 10:46:22AM +0100, Branko Cibej wrote: Mo DeJong wrote: I reran autoconf, did a ./config.status --recheck ; ./config.status and now the problem seems to have fixed itself. You shouldn't be running autodonf directly. Instead, you should run the buildconf script in the top-level dir. Just try that, without all the patches; I think your problems will just magically go away then. Yes... and rerun configure from scratch, rather than a weasely little recheck :-) Note that SVN creates a file called config.nice at the top level. If you run that, it will rerun configure with all of the proper switches. [ and that will also recurse into APR and Neon to reconfig them ] Cheers, -g -- Greg Stein, http://www.lyra.org/
Re: ARGH!
On Wed, Dec 13, 2000 at 06:24:25AM -0800, [EMAIL PROTECTED] wrote: | | The CVS re-org is going to make this alpha take a bit more time than usual | (as if it hasn't already taken long enough). Most of the instructions | refer to files and directories that may not be there any more. I just downloaded and ./configure'd the a9 tarball. It seems that the various test directories (test, srclib/apr/test, srclib/apr-util/test) didnt get populated, and this causes the make to bomb. Those should have been completely removed :-( That's what the rolling a tarball page says to do. I bet we are trying to build test by default, which is just wrong. We attempt to recurse into test in Apache and apr-util to clean them up. Neither directory builds anything by default. [ we discussed this aspect last week ] I had no idea that we didn't even ship them, though! That doesn't feel quite right to me. No opinion on Apache, but on APRUTIL, I'd like to see that testdbm gets shipped because it is actually a nice little tool for creating and inspecting *DBM files. Cheers, -g -- Greg Stein, http://www.lyra.org/
Fw: Problems compiling apr on Linux
Looks like it's fixed. Maybe we need better instructions? Maybe we don't. david On Wed, 13 Dec 2000, David Reid wrote: Hmm, I don't think you should be using that version of getpass? What is HAVE_GETPASS set to in apr_private.h. I think as you probably have an os getpass this should be set to 1. david You are right, that var was not set. In fact, the whole file was empty. I reran autoheader and autoconf and now it compiled. Thanks. Mo
RE: ARGH!
From: Greg Stein [mailto:[EMAIL PROTECTED] Sent: Wednesday, December 13, 2000 4:01 PM I had no idea that we didn't even ship them, though! That doesn't feel quite right to me. No opinion on Apache, but on APRUTIL, I'd like to see that testdbm gets shipped because it is actually a nice little tool for creating and inspecting *DBM files. nak ... we have always had a policy of pulling all test/ directories from any tarball or binaries build. However, I don't see an issue with creating a util/ directory that has thing like you describe - small bits of useful code that help the user with the package. Or just call it support/ in keeping with httpd's structure.