Hello Paul, all,
http://thread.gmane.org/gmane.comp.sysutils.autoconf.bugs/5266/focus=5549
* Paul Eggert wrote on Thu, Oct 12, 2006 at 09:45:24PM CEST:
Ralf Wildenhues [EMAIL PROTECTED] writes:
This patch kills $as_executable_p. This breaks libtool.m4 from
Libtool-1.5.22 (and possibly
Applied to HEAD and branch-1-5, respectively.
Cheers,
Ralf
HEAD:
2007-02-11 Ralf Wildenhues [EMAIL PROTECTED]
* doc/libtool.texi: Update copyright years.
Index: doc/libtool.texi
===
RCS file:
This patch fixes a documentation regression: It adds documentation for
the branch-1-5 macro naming scheme to HEAD's manual, listing most of the
old macro names as deprecated. Since however some of the old macro
names provide functionality not possible with LT_INIT, it also
resurrects some macros
Separating out func_mode_link into a separate file shaves off roughly
30% of the script execution time of `libtool --mode=compile'. But
having yet another script hampers portability and simplicity. A nice
thing is that it's not necessary. It's sufficient to merely not to
have to parse the bulk
This improves script time for compile mode by roughly 7% with bash on
GNU/Linux, for C++ sources, without passing --tag=CXX, i.e., letting
func_infer_tag do its job. With this patch, passing --tag=CXX saves
another 4% on top, as it should be. ;-)
This patch conflicts (trivially) with the other
On Feb 11, 2007, at 11:59 PM, Ralf Wildenhues wrote:
OK to apply? This set of patches is a helper for the 2.0 release in
that it helps test Libtool more efficiently. ;-)
Just gave it a quick read. Looks fine to me, please apply (and thank
you!).
Peter
On Feb 12, 2007, at 12:00 AM, Ralf Wildenhues wrote:
This improves script time for compile mode by roughly 7% with bash on
GNU/Linux, for C++ sources, without passing --tag=CXX, i.e., letting
func_infer_tag do its job. With this patch, passing --tag=CXX saves
another 4% on top, as it should
On Feb 12, 2007, at 12:01 AM, Ralf Wildenhues wrote:
This improves script time for compile mode by roughly 3% with bash on
GNU/Linux. Rewriting lists of objects is best done the way we do it
already with sed, that's why not all instances of $lo2o are killed.
This patch conflicts trivially
Hi Peter,
* Peter O'Gorman wrote on Sun, Feb 11, 2007 at 04:26:54PM CET:
On Feb 11, 2007, at 11:59 PM, Ralf Wildenhues wrote:
OK to apply? This set of patches is a helper for the 2.0 release in
that it helps test Libtool more efficiently. ;-)
Just gave it a quick read. Looks fine to me,
The new cmdline_wrap.test fails on most systems. For those not so
familiar with this relatively new test: what it does is take the libtool
script, set max_cmd_len very low so that all kinds of command line
wrapping actions are triggered in link mode. And then all those tests
in the testsuite
On Sun, 11 Feb 2007, Peter O'Gorman wrote:
Hi Ralf,
Could we test if test -x works and use that? I know that this is barely used,
but it bugs me that test -f does not test for the executable bit :)
I think autoconf has this:
if test -x / /dev/null 21; then
as_executable_p='test -x'
else
* Ralf Wildenhues wrote on Sun, Feb 11, 2007 at 05:47:16PM CET:
I would like to fix all failures at once, due to the intertwined code
in ltmain. Currently, I'm about half-way through. I'll post what I
have in a followup mail FYI, but won't apply it.
This patch makes the cmdline_wrap test
On Sun, 11 Feb 2007, Ralf Wildenhues wrote:
* Ralf Wildenhues wrote on Sun, Feb 11, 2007 at 05:47:16PM CET:
I would like to fix all failures at once, due to the intertwined code
in ltmain. Currently, I'm about half-way through. I'll post what I
have in a followup mail FYI, but won't apply
13 matches
Mail list logo