Created attachment 1087 [details]
This cleans up the code a bit. Compiler warnings, unused variables, ... A lot of cosmetic stuff that has no effect for the end user.
This works for me, but please take a look at it before applying.
This are still a lot for warnings left, will looks at that after this has been applied.
This one doesn't make sense to me, as color is used whatever the GTK+ version is, a contrario to cmap, success and i which are used when GTK+ version differs from 2.24.0. Won't compile if GTK+ version is 2.24.0.
--- src/compose.c 2011-12-30 06:00:40.000000000 +0100
+++ src/compose.c 2012-02-28 11:35:23.126835114 +0100
@@ -801,8 +801,8 @@ static void compose_create_tags(GtkTextV
buffer = gtk_text_view_get_buffer(text);
What's your GCC and GTK+ versions? It may explain some hunks in your patch!
Also, I think that removing src/gtk/headers.h will break translations.
Re comment #3: yes, it will.
This is an Arch Linux system.
Without the change to src/compose.c I got:
Nice catch for #define ICONS 21! :-)
Oh, yes you're right for color being unused in such case, sorry. Since it's not static, it's really set for nothing or I missed something.
- c = read(source, &buf, 1);
+ read(source, &buf, 1);
These are there to silence warnings about non-used return values. Of course, now there are unused variables warnings. The best would be to actually check everything went fine.
My gcc is happy without checking for return values. :D
Ok, so all of these should look like this?
if (read(source, &buf, 1) < 0)
g_warning("count not read.\n");
Yep. Actually, use debug_print(), because these really shouldn't fail - otherwise they would be checked :)
Created attachment 1088 [details]
Removed everything related to headers.h (though gcc complains about unused 'HEADERS' now) and added error handling. Hope I got everything right.
Created attachment 1092 [details]
Rebases on current cvs...
Last activity was two and a half week ago... Still any objections or will this be merged?
Christian, why don't you suggest the patch again, this time against the current CVS and let's see if we can get the developers interested? :)
Oh, I see a lot of "Reversed (or previously applied) patch detected!" messages... When did that happen?
I will try to find some time to rebase the patch.
Created attachment 1193 [details]
Ok, most of this as been applies or been dealt with. I have created a new patch against 3.9.0cvs18 with what remains.
Thanks for taking the time to update the patch. Just a few things:
for src/gtk/authors.h, the list is alphebetized (by last name)
I don't understand the need for the GtkAdjustment/GtkObject replacements either? There are no warnings as the code is. (There may have been at the time of first patch submittal though)
Created attachment 1194 [details]
I am pretty sure I had some warnings about GtkAdjustment being deprecated... But to be honest: I can not find any documentation about that.
Ok, updated again:
* fixed alphabetical order in src/gtk/authors.h
* removed everything about GtkAdjustment
* added a real fix for gnutls
I have build errors with the patch applied though. Mostly the removal of r and error cause problem as it seems we're now using these variables :)
I've applied the AUTHORS part anyway as it seems we forgot to add you in the beginning!
Created attachment 1195 [details]
Ok, last patch had some problems... Back in march some variables were unused but are no longer in latest version...
Additionally my fix for gnutls was wrong. We should keep it this way if we want to be compatible with gnutls <= 2.10.
But I changed type to not use the deprecated ones from gnutls 1.x.
So here is an updated version.
There are some more glib and gtk warnings, but I will look at these if we got this right.
Changes related to this bug have been committed.
Please check latest CVS and update the bug accordingly.
You can also get the patch from:
2012-11-23 [colin] 3.9.0cvs20
Cleanup some warnings and deprecated types. Patch by
Christian Hesse, fixing bug #2617. Thanks!