libc dependency in 0.8.0pre (bug 487)
Folks, Updated my CVS at home (Debian 3.0) to 0.8.0pre the other day, and I like what I see. When I tried to do the same at work today (Redhat 6.2), I had problems building. Dpid uses mkdtemp() which was introduced in glibc 2.1.9. However, ./configure has no problems with my glibc 2.1.3. Submitted as bug 487. This constitutes a change from the 0.7 days when (according to the Compatibility web page) any glibc2 would do. Introducing (by hand) a definition of mkdtemp() I found on the web solved all my problems. Browsing the source suggests that you are not in the habit of carrying around lot of little libc bits---and the associated preprocessor code munging---for systems with variant or insufficient libc's. That is fine. It keeps the code base small and simple, but perhaps ./configure should look out for these things. So does this mean some previously working libc's are now going to be excluded, that more conditional compilation stuff will need to go in, or that dpid needs some work to increase it's universal compatibility? -- -- David McKee -- dmckee@jlab.org -- (757) 269-7492 (Office)
David, On Tue, 12 Aug 2003, David McKee wrote:
Folks, Updated my CVS at home (Debian 3.0) to 0.8.0pre the other day, and I like what I see. When I tried to do the same at work today (Redhat 6.2), I had problems building. Dpid uses mkdtemp() which was introduced in glibc 2.1.9. However, ./configure has no problems with my glibc 2.1.3.
Submitted as bug 487.
This constitutes a change from the 0.7 days when (according to the Compatibility web page) any glibc2 would do.
Introducing (by hand) a definition of mkdtemp() I found on the web solved all my problems.
Browsing the source suggests that you are not in the habit of carrying around lot of little libc bits---and the associated preprocessor code munging---for systems with variant or insufficient libc's. That is fine. It keeps the code base small and simple, but perhaps ./configure should look out for these things.
So does this mean some previously working libc's are now going to be excluded, that more conditional compilation stuff will need to go in, or that dpid needs some work to increase it's universal compatibility?
Don't panic! The mkdtemp() issue needs to be solved. Thanks for reporting the problem. If configure has a way to deal with it, we'd need to find it out how. If not, just to think of something else (as implementing a custom function as you did, or see if there's a wrapper in glib etc.) I'm open to suggestions. Cheers Jorge.-
On Wed, Aug 20, 2003 at 01:02:40PM -0400, Jorge Arellano Cid wrote: [snip]
Don't panic!
The mkdtemp() issue needs to be solved. Thanks for reporting the problem.
If configure has a way to deal with it, we'd need to find it out how. If not, just to think of something else (as implementing a custom function as you did, or see if there's a wrapper in glib etc.)
I'm open to suggestions.
Well, you could add a portable implementation e.g. from the *BSD libc to dillo and a AC_CHECK_FUNCS to autoconf to conditionally include that code. This is the way coreutils and other GNU packages work. I'll send you a working copy if you want. Joerg
Cheers Jorge.-
_______________________________________________ Dillo-dev mailing list Dillo-dev@lists.auriga.wearlab.de http://lists.auriga.wearlab.de/cgi-bin/mailman/listinfo/dillo-dev
On Thu, 21 Aug 2003, Joerg Sonnenberger wrote:
On Wed, Aug 20, 2003 at 01:02:40PM -0400, Jorge Arellano Cid wrote: [snip]
Don't panic!
The mkdtemp() issue needs to be solved. Thanks for reporting the problem.
If configure has a way to deal with it, we'd need to find it out how. If not, just to think of something else (as implementing a custom function as you did, or see if there's a wrapper in glib etc.)
I'm open to suggestions.
Well, you could add a portable implementation e.g. from the *BSD libc to dillo and a AC_CHECK_FUNCS to autoconf to conditionally include that code. This is the way coreutils and other GNU packages work.
I'll send you a working copy if you want.
Of course! It'd be of great help. Cheers Jorge.-
On Tue, Aug 12, 2003 at 04:33:43PM -0400, David McKee wrote:
Folks, Updated my CVS at home (Debian 3.0) to 0.8.0pre the other day, and I like what I see. When I tried to do the same at work today (Redhat 6.2), I had problems building. Dpid uses mkdtemp() which was introduced in glibc 2.1.9. However, ./configure has no problems with my glibc 2.1.3.
Submitted as bug 487
Could you try the attached patch and report success or failure? Esp. whether the included mkdtemp.c compiles clean and dpiddir.c gets the prototype. Thanks, Joerg
participants (3)
-
David McKee -
Joerg Sonnenberger -
Jorge Arellano Cid