On Fri, Apr 9, 2010 at 7:05 PM, Malthe Borch <mbo...@gmail.com> wrote: > I think it would be worth it to go through the generated code for > Plone to see if it looks optimal, or if there are obvious problems > such as the one you've pointed out. Rinse and repeat for every major > change, perhaps.
Well, the current generated code has deteriorated quite a bit with features being added over time ;) Last time I looked there were a number of obvious inefficiencies: - Some page template classes always have certain variables in their expression context. Like context, view, root or the like. Currently there's a lot of additional attribute accesses to get these from the general expression context. - The utf-8 / Unicode mixed mode generates lots of additional code. Only allowing Unicode or objects that have an Unicode representation would simplify the code. The __html__ feature could be optional as well. - Generally there's no real optimizations in the code and the current infrastructure makes them hard to write. Like generating efficient code for stuff like "var1 string:'static'". Or not filling the repeat variable if it isn't used inside a repeat block. Maybe I should try again harder, but I failed at both of these the last times I tried ;) Hanno _______________________________________________ Repoze-dev mailing list Repoze-dev@lists.repoze.org http://lists.repoze.org/listinfo/repoze-dev