Peter O'Gorman wrote: > Ralf Wildenhues wrote: > | lt_dlcaller_register takes zero args in branch-1-5, 2 in HEAD. > | This sucks. CVS HEAD m4 wants two args, but does not test for this > | (and thus loves to use an installed ltdl from branch-1-5 over its own > | convenience one from HEAD). > | > | We need to kill this in order to avoid being backward incompatible. > | I can only hope no packages other than m4 depend on this already. :( > > Looked at this (as it is a showstopper) and agree. Added api (perhaps > lt_dlcaller_set_iface(const char*,lt_dlhandle_interface *); or > something) to > set the info and remove the arguments here. > > Gary?
So, I think the real problem is that LT_WITH_LTDL from libtool-2.0 will need to check for a libltdl that provides the lt_dlhandle_first function to make sure libtool-2.0 using apps get the libtool-2.0 version of libltdl... And that falls in the remit of the LT_WITH_LTDL rewrite I'm working on at the moment. Gah! :o) Cheers, Gary. -- Gary V. Vaughan ())_. [EMAIL PROTECTED],gnu.org} Research Scientist ( '/ http://tkd.kicks-ass.net GNU Hacker / )= http://www.gnu.org/software/libtool Technical Author `(_~)_ http://sources.redhat.com/autobook
signature.asc
Description: OpenPGP digital signature