Steve Youngs <steve <at> sxemacs.org> writes:
> Excellent! I'm not sure if we have any Solaris blood here. Would you
> like to join us? Even if the only things you are able to contribute
> are the occasional test build and healthy bug report, I'm sure it'll
> add a lot to SXEmacs.
>
I'm hesitant to make any more commitments at this time, but I'll consider it.
> > First, it seems like you need the following patch:
> > http://list-archive.xemacs.org/xemacs-patches/200503/msg00074.html
>
> Ignoring the actual patch for a moment, this comment at the top of
> that email...
>
> The !defined(HAVE_DECL_SYS_SIGLIST needs to stay so that systems not
> using configure (i.e. Windows) don't barf.
>
> Wouldn't that imply that we _DON'T_ need the patch? (SXEmacs doesn't
> support windoze) Also, XE 21.4 doesn't have this change. That tells
> me that either Vin hasn't gotten around to applying it yet, or it
> isn't needed in 21.4 (which means it probably isn't needed in SXEmacs).
>
> Anybody care to go out on a limb and say whether or not we need this
> patch (reproduced here for convenience): ?
>
The comment prior to the patch is a bit of a puzzle. However, the ChangeLog
comment in the patch (which you did not reproduce) is more telling:
--- src/ChangeLog 2005/03/11 19:23:37 1.809
+++ src/ChangeLog 2005/03/12 04:15:25
@@ -1,3 +1,9 @@
+
+2005-03-11 David Robinow <drobinow@xxxxxxxxx>
+
+ * syssignal.h: Autoconf always sets HAVE_DECL_SYS_SIGLIST. Test
+ for non-zero.
+
2005-03-11 Stephen J. Turnbull <stephen@xxxxxxxxxx>
* XEmacs 21.5.20 "cilantro" is released.
> Index: src/syssignal.h
> ===================================================================
> RCS file: /pack/xemacscvs/XEmacs/xemacs/src/syssignal.h,v
> retrieving revision 1.16
> diff -u -r1.16 syssignal.h
> --- src/syssignal.h 2005/03/11 11:20:18 1.16
> +++ src/syssignal.h 2005/03/12 04:15:27
> <at> <at> -270,7 +270,7 <at> <at>
> /* HAVE_DECL_SYS_SIGLIST is determined by configure. On Linux, it
> seems,
> configure incorrectly fails to find it, so s/linux.h defines
> HAVE_SYS_SIGLIST. */
> -#if !defined (HAVE_DECL_SYS_SIGLIST) && !defined (HAVE_SYS_SIGLIST)
> +#if (!defined(HAVE_DECL_SYS_SIGLIST) || !HAVE_DECL_SYS_SIGLIST ) && !defined
(HAVE_SYS_SIGLIST)
> extern const char *sys_siglist[];
> #endif
>
> > But, then I get some other errors and make fails:
> > ...
> > gcc-3.3.4 -Wall -Wno-switch -Wmissing-prototypes -Wsign-compare
> > -Wno-char-subscripts -Wundef -Wstrict-prototypes -Wpacked -Wpointer-arith
> > -Wshadow -Wmissing-declarations -L/opt/sfw/lib -L/opt/sfw/kde/lib
>
> Just an idle comment, I don't think this is related to you prob... kde
> libs? Why oh why is the kde lib directory referenced there?
>
Sorry, my bad. We mainly use Motif/CDE and I just recently installed Gnome and
KDE. Thanks for the heads up!
> [...]
>
> > EmacsManager.o ../lwlib/liblw.a -lXm -ltiff -lpng -ljpeg -lz -lXpm
> -ltt
>
> -ltt... that'd be tooltalk, right?
>
> > Undefined first referenced
> > symbol in file
> > trunc effi.o
> > vars_of_tooltalk emacs.o
> > syms_of_tooltalk emacs.o
>
> Hmm, definitely looks like tooltalk is at least partially
> responsible. Have a look in `Installation' in the root of your build
> directory, does it mention tooltalk being enabled? If yes, try
> turning it off (./configure --without-tooltalk), if no, try turning it
> on.
>
> > error_check_ffiobject effi.o
>
> There seems to be other things wrong too. Try turning off FFI
> (--disable-ffi).
>
> Lastly, but most importantly, could you please submit this bug to our
> issue tracker at: http://issues.sxemacs.org/ Attach your
> `Installation' file, the relevant make output, and anything else you
> feel might be relevant.
>
> Thank you very much for your feedback.
>
They are both there:
Compiling in support for ToolTalk.
Compiling in support for FFI.
What confused me is that they are both autodetected, but I guess that doesn't
mean that they will work. I'll wait to see if there are any other comments
before making this into an official bug report.
Thanks,
Rodney
|