- Friday, September 19, 2014

openafs@conference.openafs.org

Friday, September 19, 2014< ^ >

Room Configuration

Room Occupants

GMT+0
[00:31:47] meffie leaves the room
[09:00:02] Jeffrey Altman leaves the room
[09:00:51] Jeffrey Altman joins the room
[10:02:49] Simon Wilkinson joins the room
[10:42:55] Simon Wilkinson joins the room
[10:44:00] Simon Wilkinson leaves the room
[10:47:09] Simon Wilkinson joins the room
[10:48:14] Simon Wilkinson leaves the room
[10:57:21] Simon Wilkinson joins the room
[10:58:25] Simon Wilkinson leaves the room
[11:08:18] Simon Wilkinson joins the room
[11:09:23] Simon Wilkinson leaves the room
[11:11:31] Simon Wilkinson joins the room
[11:12:36] Simon Wilkinson leaves the room
[11:14:15] Simon Wilkinson joins the room
[11:15:19] Simon Wilkinson leaves the room
[11:56:51] meffie joins the room
[12:35:25] Simon Wilkinson leaves the room
[14:09:48] ballbery leaves the room
[14:09:49] ballbery joins the room
[14:42:21] meffie leaves the room
[15:00:24] meffie joins the room
[16:29:23] mvita leaves the room
[16:29:24] mvita joins the room
[16:29:47] mvita leaves the room
[16:45:37] ktdreyer leaves the room
[16:46:31] ktdreyer joins the room
[18:45:37] ballbery leaves the room
[18:45:38] ballbery joins the room
[19:39:17] kaduk joins the room
[19:39:54] <kaduk> Are there reasons to use one of top_builddir vs. TOP_OBJDIR in any particular places?
[20:03:35] <kaduk> (In particular, setting top_builddir in Makefile.config is somewhat questionable, as it gives any Makefile including Makefile.config the value appropriate for src/config, which is not necessarily appropriate for all build directories.)
[20:52:31] meffie leaves the room
[20:54:39] ballbery leaves the room
[21:17:44] <jhutz@jis.mit.edu/owl> top_builddir should not differ between directories.
[21:18:00] <kaduk> It's expressed as a relative path.
[21:18:22] <kaduk> If you have configure substitute @top_builddir@ in any given Makefile, it is correct, but will be different in different levels of the hierarchy.
[21:18:41] ballbery joins the room
[21:19:03] <jhutz@jis.mit.edu/owl> True.  I thought top_srcdir et al were substituted in every Makefile.
[21:19:31] <jhutz@jis.mit.edu/owl> Of course, I've been fighting with this build system since way before
autoconf, so I may be mixing memories from different eras.
[21:21:07] <kaduk> Maybe.
[21:21:26] <kaduk> In the present state there's just one main place, in Makefile.config, and Makefile.config gets included in ~every other Makefile.
[21:21:53] <kaduk> You wouldn't happen to know offhand what syntax libool -export-symbo-regex uses to represent logical ORing, do you?
[21:25:42] <kaduk> Hmm, maybe I was just getting bitten by make eating the '$'s. *performs further experimentation*
[21:25:46] <jhutz@jis.mit.edu/owl> No, but portability suggests it can only be one of the two POSIX RE
syntaxes.  So either it's | or you lose.
[21:29:42] <kaduk> Hmm, it seems to be working now.
[21:42:16] <kaduk> Yeah, so libtool -export-symbols is an exact list of the symbols which are to be exported by the library.  If you don't pass -no-undefined, you're allowed to have those not be present when linking the actual library, but they're still exported from the library.  Which means that there is an implicit assumption that when you go and use that library, you will be providing some other library that provides those missing symbols.
[21:42:32] <kaduk> Thus, our LT_LDLIB_missing is actually useless.
[21:43:03] <kaduk> We can, however, use ugly shell tricks to turn a list of symbols that we should export if they are present, into a regex and use -export-symbols-regex.
[22:23:47] kaduk leaves the room
[22:30:07] ballbery leaves the room