Thunderbird projects and roadmaps

Robert Kaiser KaiRo at KaiRo.at
Wed Sep 12 12:21:45 UTC 2012


Joshua Cranmer schrieb:
>   * Recent discussions in m.d.planning appear to indicate that the
>     current process is not stopping significant performance regressions
>     from happening.

Actually, the really huge ones get detected just fine, the problem is 
the ones up to 5% or so, as the Talos perf tests are somewhat noisy and 
therefore it's (depending on which Talos test you're looking at) 
sometimes hard to distinguish signal from noise when it comes to smaller 
regressions.

In any case, I think the main issue for Thunderbird is probably 
"snappiness", i.e. us blocking the main thread and therefore the UI too 
often and too long. Things like updating newsgroups or feeds surely are 
a factor there (should really run in the background in some way instead 
of blocking the main thread), as well as probably some other actions, 
possibly including some stuff we do at startup that we could do lazily 
and/or async.

Robert Kaiser



More information about the tb-planning mailing list