Re: [evolution-patches] fix for bug #33933 (message-list sorting)

[Not Zed <notzed ximian com>]

Oh, that is bad.  It does make a major difference.

Umm, wouldn't it be like 10x easier just to add the 5 lines of code to reset the normalised_string tables when set_folder() is called??

On Thu, 2005-01-27 at 09:11 -0500, Jeffrey Stedfast wrote:
> I think it does affect it (seems to for me but I didn't do any actual
> timing).
>
> We'll probably have to come up with a different hack to make it faster
> at some point, but the mechanism this patch removes is no longer viable.
>
> Jeff
>
> On Thu, 2005-01-27 at 10:48 +0800, Not Zed wrote:
> > 
> > Does this affect sorting performance, or does etable implicitly cache
> > such sorting?
> > 
> > 
> > On Tue, 2005-01-25 at 15:31 -0500, Jeffrey Stedfast wrote:
> > > This should probably go into 2.0.x as well...
> > > 
> > > The problem was that back in 1.4 when we sued different MessageList
> > > objects per folder, I had implemented a sorting optimisation that
> > > cached
> > > the collated keys for each subject string (and sender/to/etc - which
> > > is
> > > odd because the sorter didn't actually use those, but ah well).
> > > 
> > > Anyways, now that we reuse the same "view", this hack no longer
> > > works
> > > and in fact causes problems (as seen in the report).