What's the plan and timetable to fix Thunderbird performance at startup due to Lightning bugs and flaws

ISHIKAWA,chiaki ishikawa at yk.rim.or.jp
Thu Jul 18 21:57:27 UTC 2019


In https://bugzilla.mozilla.org/show_bug.cgi?id=1502923#c33
you listed a few issues related to network download of data.

Is it possible that TB uses unbuffered I/O for these network I/O?

I have been trying to fix pop3 (and imap) to use buffered write. 
Currently it does not.
(I am re-creating the patches to fit the modified source files over the 
last serveral months.)
The delay may not be noticeable if all your data is on local disk.
However, once it is on a network file system, we lose badly due to the 
large overhead of writing in a network file system.
We definitely should use buffered write for this.

Just my two cents worth.

Chiaki


On 2019/07/18 17:42, Richard LEGER wrote:
> Dear Thunderbird (TB) team,
>
> Startup performance issues have been plaguing TB for years (7 years or 
> more possibly) especially due to Lightning being enabled... and it 
> became worth in past years...
>
> Up to now the "false" assumption has been that all I/O happens in one 
> thread (ui, email core features in addition of Lightning) causing too 
> much processing... not responding issues...
>
> While this may have been partially true, tremendous progress have been 
> made so far in that regards within TB (e.g core improvements, use of 
> workers and async features helping if I am not mistaking), so it had 
> been more an aggravation of existing underlying issues in Lightning 
> than a root cause.
>
> Indeed I think such progress shadows the fact that Lightning still 
> contains bugs and design flaws (no blame on developers here) that are 
> mainly source of performance issues especially at startup that are 
> still not being analysed and addressed the way they should (at least 
> from end-users point of view) by implementing long term 
> fixes/solutions into that code so it would performs "normally" ;-)
>
> As an end-user, month ago, I have revealed (because end-users could 
> not take it any more!) bug 
> https://bugzilla.mozilla.org/show_bug.cgi?id=1502923 (patch issued but 
> backed out without further support) and bug 
> https://bugzilla.mozilla.org/show_bug.cgi?id=1543953 as the tip of the 
> iceberg but received barely any support from TB/Lightning dev teams so 
> far or feedback on helping resolving those long lasting and 
> highlighted performance issues. It may not be a priority for 
> developers (in their time and resources constraints) but it IS for 
> many end-users...
> [Update: there might be light at the end of the tunnel as per new bug 
> https://bugzilla.mozilla.org/show_bug.cgi?id=1567055 created today]
>
> From my impression, while it seems that lot of efforts (and 
> communication) are oriented new UI revamping and new features... bugs 
> fixing and especially those related to performance (quality, 
> reliability, speed) seems relegated to a lower priority to some 
> extent... while it should be in reverse order...
>
> Currently are code reviews and new features subject to large scaling 
> tests to assess performance of code with large data sets prior 
> publishing it?
> Are performance issues detected (or reported) fixed immediately or 
> right after in next dev cycles?
>
> For two years TB quality and reliability had decreased drastically, 
> mainly due to lengthy discussion about governance that put TB 
> development and maintenance to a stall, it was kept afloat by the 
> remaining dev team and community (great thanks to them!).
>
> As it is now past, shouldn't 2019 be the year where TB reliability and 
> performance are brought back to life and take priority over new 
> features? Wouldn't that be the best way to bring back confidence (and 
> contributors) in the project? Wasn't it the priority for 2019 "Making 
> Thunderbird fly faster" in the first place, as I recall?
>
> Now that a new team is up and running, what is the plan to tackle 
> reliability and performance issues in TB especially those linked to 
> Lightning?
>
> Would (and could) time and resources be dedicated to them in 
> particular so they can finally be fixed once and for all in the next 
> few months?
>
> While Lightning is an add-on it provide a calendar features to TB and 
> therefore shall be considered as an essential Core feature that need 
> much care and attention ;-) I am not saying it is not already (because 
> it certainly is)... just that it needs much more... especially 
> performance wise...
>
> As of today it is still not possible to have 4-5 calDav calendars 
> (with thousands of items each) enabled and active at the same time, 
> above two, it cripples TB especially at startup...
>
> Maybe a plan and timetable is already in place if that the case, 
> please advise and communicate on what can be expected...
> Otherwise could a plan and timetable be established with clear goals 
> and results to clear any performance issues in TB due to Lightning 
> especially, as a priority?
>
> It may be a wishful thinking but plan, support and fixes on the 
> performance matters would be greatly appreciated by end-users...
>
> Looking forward to seeing Thunderbird fly fast in 2019!
>
> Regards,
> Richard
>
> _______________________________________________
> tb-planning mailing list
> tb-planning at mozilla.org
> https://mail.mozilla.org/listinfo/tb-planning




More information about the tb-planning mailing list