Re: implementation of message threadings.



2000-11-06 05:34, Atsuhiko Yamanaka rašė:
> This mail is attached a patch file to balsa-1.0.pre5,
> which includes message threading functions and
> some hacks for improving the performance of balsa.
> 
> I wrote two message threading functions and they are
> included src/balsa-index-threading.c 
> 
> And I had to do some hacks for improving the performance of balsa,
> because the message threading functionality heavily depends  on the
> GtkCTree widget and it is not so lightweight in comparison with the GtkCList.
> Try following operations on balsa-1.0.pre5 and patched balsa-1.0.pre5.
>   * select hundreds messages and transfer to other mailbox.
>   * open a mailbox, which includes hundreds messages and exit(type C-q).
> If you use balsa on the remote host, you will find improvements clearly.

I have tested this patch. Of course, threaded message view is very nice, I
like it very much. One of a few reasons I had to run mutt was to easily read
big threads.

However, on some mailboxes I encountered a crash. Debugging in gdb, I
have tracked that a SIGSEGV is got in gen_container function from
balsa-index-threading.c file, gdb writes this:

Program received signal SIGSEGV, Segmentation fault.
0x40486b38 in g_str_hash () from /usr/lib/libglib-1.2.so.0

I attach one small mailbox, which causes this.

OTOH, those speed improvements look fine. As I understand, they just make
some operations on some messages at once, not one by one. And more... But no
objections about this, I think parts of the patch can be integrated into
1.0.

Note to Atsuhiko: maybe you could split a patch into two: 1) improvements 2)
threading? Also, never include Makefile.in and similar generated files in a
patch. They are large but useless.

-- 
Gediminas Paulauskas    -=-  http://03bar.ktu.lt/~menesis/
El. paštas lietuviškai  -=-  http://www.liks.lt/elpaliet.htm

ktug



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]