Hello Paolo, Akim, * Paolo Bonzini wrote on Fri, Nov 28, 2008 at 04:41:23PM CET: > Akim Demaille wrote: > > Do you really mean --recheck? In that case there is no reason for > > config.status to recreate libtool, unless I missing something. Don't > > you mean "./config.status libtool"? > > Seeing that LIBTOOL_DEPS is "$ltmain", yes. I pushed the obvious fix.
> --- a/doc/libtool.texi > +++ b/doc/libtool.texi > @@ -1986,6 +1986,6 @@ @defmacx AM_PROG_LIBTOOL > @example > LIBTOOL_DEPS = @@LIBTOOL_DEPS@@ > libtool: $(LIBTOOL_DEPS) > - $(SHELL) ./config.status --recheck > + $(SHELL) ./config.status libtool Actually, the documentation used to be correct for Libtool 1.5.x, because there, the libtool script was really generated by configure and not by config.status. Things have changed in 2.2.x: libtool is now normally generated by config.status. However, if LT_OUTPUT is used, then it causes libtool to already be created by configure (and then be re-created, hopefully identically, by config.status). This macro was added for those packages which need to use the libtool script in configure tests. I haven't checked yet whether it is possible to forego the --recheck part each time. Also, tests/old-m4-iface.at still has the old-style code. I did like being able to use the 2.2.x test suite on an 1.5.x script with make check-local TESTSUITEFLAGS=LIBTOOL=/path/to/libtool-1.5.x/libtool but over time, that will be less and less important, so I wonder if that should be fixed as well. Anyway the situation now seems inconsistent. Cheers, Ralf
