Re: libtool replace -Xlinker by -Wl can lead to problems

2003-09-20 Thread Peter O'Gorman
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Boehne, Robert wrote: | Alternatively, I think gcc/g++/g77/gcj can use -Xlinker instead of -Wl. This seems reasonable, perhaps Nicolas would be willing to do a test first? Nicolas, after running configure can you edit the generated libtool script and

Re: Link performance: some numbers and a hack

2003-09-20 Thread Peter O'Gorman
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Iñaki García Etxebarria wrote: |>Patches which improve speed and cause no harm will be accepted. | | I will be glad to write a patch (btw, any place to get libtool-1.5a | tarballs?), but since I know quite little about linking outside the | marvelous wo

RE: Link performance: some numbers and a hack

2003-09-20 Thread Bob Friesenhahn
On Fri, 19 Sep 2003, Iñaki García Etxebarria wrote: > > The information in the dynamic libs may be incorrect, or it may not > > actually exist. > Uff. Does this really happen out there? > > > It would be good to see why linking is so slow when all the libraries > > are listed. Is the same library

RE: Link performance: some numbers and a hack

2003-09-20 Thread Iñaki García Etxebarria
> > It's libtool which is slow, the actual linking is pretty fast once gcc > > is started. Say, libtool takes around 2 minutes and gcc 2 seconds. > > That is pretty atrocious. My point exactly ;-) > This link line is what is passed to libtool. What is of interest is > the link line which is bein