SQL backend performance

[Derek Atkins]

Donald Allen <donaldcallen at gmail.com> writes:

> Some good news:
>
> Doing this the easy way first, I did a little manual pc sampling. I
> ran gnucash (today's trunk) under gdb, let it get to the point where
> it begins to load my data from postgresql, and periodically ctrl-c'd
> in gdb and copied the interrupted location and a backtrace to an emacs
> buffer. In the interest of saving space in this message, rather than
> just copying the result into the message (I assume that attachments
> are not permitted in emails to the gnucash mailing lists), I'll tell
> you that I interrupted execution 7 times and the first 6 were
> identical to this:
>
> 0x00007ffff0471187 in g_list_append () from /usr/lib/libglib-2.0.so.0
> (gdb) bt
> #0  0x00007ffff0471187 in g_list_append () from /usr/lib/libglib-2.0.so.0
[snip]

> (gdb) bt
> #0  0x00007ffff0471187 in g_list_append () from /usr/lib/libglib-2.0.so.0
> #1  0x00007fffe86d55c0 in load_splits_for_tx_list (be=0xa5a300,
> stmt=<value optimized out>) at gnc-transaction-sql.c:238
[snip]

Yeah, we should almost NEVER use g_list_append!!  I still see 206
references to g_list_append in the source tree.  We might want to take a
look at all of those to see if they are ever handling lists of size > 10.

-derek
-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord at MIT.EDU                        PGP key available