Hi guys! Did you test 0.7.3 as I suggested? The good/bad news (depends on being optimistic/pesimistic!) is that yesterday I found a bug. The newest CVS has a g_print that makes it evident. Valgrind says: ==983== Invalid memory access of size 1 ==983== at 0x4014861B: strcmp (in /usr/local/lib/valgrind/valgrind.so) ==983== by 0x403CABAD: g_str_equal (in /usr/lib/libglib-1.2.so.0.0.10) ==983== by 0x403BDB38: g_hash_table_lookup_extended (in /usr/lib/libglib-1.2 ==983== by 0x805B912: p_Dw_gtk_viewport_put_anchor (dw_gtk_viewport.c:770) ==983== Address 0x41054960 is 0 bytes inside a block of size 12 free'd ==983== at 0x40148050: free (in /usr/local/lib/valgrind/valgrind.so) ==983== by 0x403C1FE8: g_free (in /usr/lib/libglib-1.2.so.0.0.10) ==983== by 0x805D646: Dw_page_destroy (dw_page.c:338) ==983== by 0x402EB876: gtk_marshal_NONE__NONE (in /usr/lib/libgtk-1.2.so.0.9 Note the p_Dw_gtk_viewport_put_anchor where the g_print is. This may come from somewhere else, so let's hunt. It should be solved before making the release. Cheers Jorge.-
On Wed, 9 Jul 2003, Jorge Arellano Cid wrote:
[...] The good/bad news (depends on being optimistic/pesimistic!) is that yesterday I found a bug. The newest CVS has a g_print that makes it evident.
I digged in a bit more and found the bug! Just a missing "break;" inside a case. The patch is in CVS. In any case, I'll wait for Sebastian to confirm me that. At first I hesitated because 5 anchors in the splash page produced about 15 extra calls, but this seems OK because they need to update the offset as the text rewraps. Once this is confirmed, 0.7.3 will be released (on Thursday). Cheers Jorge.-
On Wed, Jul 09, Jorge Arellano Cid wrote:
On Wed, 9 Jul 2003, Jorge Arellano Cid wrote:
[...] The good/bad news (depends on being optimistic/pesimistic!) is that yesterday I found a bug. The newest CVS has a g_print that makes it evident.
I digged in a bit more and found the bug! Just a missing "break;" inside a case. The patch is in CVS.
In any case, I'll wait for Sebastian to confirm me that. At first I hesitated because 5 anchors in the splash page produced about 15 extra calls, but this seems OK because they need to update the offset as the text rewraps.
This seems correct, but I haven't yet examine it in detail. Anyway, current CVS has this g_print line commented. AFAIS, this would be the last issue delaying the release, what I see else in the bug track: - 470: Is this the one which has been fixed? - 476: This is already older (tested with 0.7.1, which happened to be available :-), so it is not something making dillo more buggy than before. Sebastian
On Wed, Jul 16, Sebastian Geerken wrote:
On Wed, Jul 09, Jorge Arellano Cid wrote: [..]
In any case, I'll wait for Sebastian to confirm me that. At first I hesitated because 5 anchors in the splash page produced about 15 extra calls, but this seems OK because they need to update the offset as the text rewraps.
This seems correct, but I haven't yet examine it in detail. Anyway, current CVS has this g_print line commented.
Now I have looked at it in detail, and this confirms my assumtion. p_Dw_gtk_viewport_put_anchor is called in Dw_page_size_allocate, which is OTOH called quite often, but (especially in the meanwhile) does not consume much time. Sebastian
participants (2)
-
Jorge Arellano Cid -
Sebastian Geerken