Hi,
Sorry, I can't get your point, but as concerned rpath issue I set
LD_LIBRARYN32_PATH (yeah, IRIX has LD_LIBRARY(N32/64)_PATH for
binary compatibility) as environment variable in advance, and
required libraries are already hard-coded run-time-library-path
in itself. So,
On Thu, 3 Feb 2005 01:01:32 +0100
Christian Biere wrote:
> Daichi Kawahata wrote:
> > Although it sounds pretty dull, IRIX's linker needs precise
> > order, i.e. local library needs to be located ahead while
> > global one needs behind and
> >
> > "gcc $(GLIB_CFLAGS) $(GLIB_LIBS) -o has_glib_func has_glib_func.c"
> >
> > will fail due to location of "$(GLIB_LIBS)", that makes link-time
> > complicated on IRIX nowadays.
>
> Doesn't it also need -R/path1:/path2 instead of -R/path1 -R/path2
> because it'll use the last -R only?
(possibly) no, I think it doesn't need in gtkg configure process,
in this connection, compiled program will search library from last
rpath, if doesn't exist then search next last path. Well, I don't
know certainly, because I never use -R flag (instead I use
-Wl,-rpath -Wl,/usr/local/lib32 as libtool wrapper).
> Well, this won't cause a problem at compile-time but the program
> won't start then. Using evil hacks from hell like setting
> LD_LIBRARY_PATH or editing /etc/ld.so.conf is a *very* *bad* idea,
> unless you know *exactly* what you're doing.
I don't know whether is problem setting LD_LIBRARYN32_PATH, I set
this colon-separated variable to avoid conflicting case (there
are /usr/lib32 has shipped libraries, /usr/freeware/lib32 from
user-land, /usr/local/lib32 has self-compiled libraries). It never
cause trouble in gtkg.
> > int main()
> > {
> > - signal(2, foo);
> > - kill(getpid(), 2);
> > + sigaction(2, foo);
> > kill(getpid(), 2);
> > printf("abc\n");
>
> I'd be surprised if that compiles at all. sigaction() takes
> completely different parameters than signal() except for the
> signal number. Again, if this check fails, nothing is lost or won.
> I've already modified the sources, so signal() isn't used anywhere
> and this check is completely redundant.
Well, can I consider the above mentioned state doesn't towards me?
> > -if $cc $ccflags $ldflags $glibcflags $glibldflags -o try try.c
> > >/dev/null 2>&1;
> > +if $cc $ccflags $ldflags $glibcflags -o try try.c $glibldflags
> > >/dev/null 2>&1;
>
> I see, thanks. I guess $ldflags should be moved to the end as well,
> right?
It doesn't need without making appearance of shell script elegant!?,
better way is to separate $flag and $lib like,
$cc $ccflags $ldflags $glibcflags $glibldflags -o try try.c $libs
then position of $flag doesn't the matter, location of actual
library is the one, in short, if I enabled NLS support the order
"-liconv -lintl" won't compile because libintl has libiconv
dependency function, but I must say again that problem never
happen in configure process, it feels like towards off-topic.
Anyway, thanks.
--
Daichi
-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
Gtk-gnutella-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel
