Re: [evolution-patches] (fwd) [patch] 64-bit cleanliness patch

[Not Zed <notzed ximian com>]

In future, attach the patch please.

On Tue, 2004-02-17 at 15:24 +1100, Martin Pool wrote:

> Could you please consider this patch?
> 
> http://bugzilla.ximian.com/show_bug.cgi?id=53164
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=198660
> 
> I don't know if it's still needed in 1.5; I'll test that soon.
> 
> Package: evolution
> Version: 1.4.5-3
> Severity: normal
> Followup-For: Bug #198660
> 
> Here is a patch to make evolution 1.4 work on ia64 (well, at least it
> starts up; that's all I know how to do... ;-).
> 
> The patch fixes the following issues:
> 
>  - configure failed for me because libdb3 has a dependency on -lpthread;
>    fixed by configure.in hacking
> 
>  - e-pilot-util.c was missing an include of <string.h> which caused strdup()
>    to be implicitly defined to return an "int", which leads to crashes on
>    64-bit platforms, because the top 32 bits will get truncated away
> 
>  - rule-context.[hc]: it was being lazy about typing and used "int" where
>    GType should have been used; again this leads to crashes on 64-bit
>    platforms because of pointer-truncation

Looks ok.

>  - mail-importer.c was missing an include of "e-util/epath.h", which lead
>    to the implicit declaration of a function returning a pointer (sorry,
>    forgot which function); this leads to the same kinds of crashes as
>    the strdup() issue explained above

This file has been moved and already fixed, and this problem is no
longer an issue.  The file is now in mail/importers/mail-importer.c
(always easiest/most accurate to build patches against CVS).

>  - mail-vfolder.c: use GTK_CONTAINER() to avoid compiler warning (I don't
>    think this patch does anything from a functionality-point of view, but
>    it does remove some annoying warning messages).

This should use a C cast, i.e. (GtkContainer *) rather than the Gtk cast
GTK_CONTAINER() for mailer code consistency.

> Looks to me like the evolution team should do occasional testing on
> 64-bit platforms.  None of these bugs are ia64-specific.

We've had plenty of 64 bit patches before.  *shrug*  When Novell buy me
a 64 bit box, i'll be more than happy to make it work on it ...