[Qgis-psc] QGIS 2.18 bug fixing effort

[Nyall Dawson]

On 23 May 2017 at 03:27, Giovanni Manghi <giovanni.manghi at gmail.com> wrote:
> Hi Nyall,
>
>
>
>>> b) 2.18.7
>>>
>>> rendering time:
>>> real    0m25.341s
>>> user    0m19.684s
>>> sys     0m1.120s
>>>

> with your patch
>
> 2.18.8
>
> rendering time:
> real    0m38.663s
> user    0m33.572s
> sys     0m1.244s
>
> so... quite slower in rendering my test 758k ploygon layer.

This patch won't affect rendering at all - if you're seeing slower
rendering after this commit, it's either:
- an earlier 2.18.8 commit. Possibly the extra safety added by
https://github.com/qgis/QGIS/commit/452c806670aa31caf0d6ec4e1f308c3d84e6d024,
but if so I'd choose slower, crash free rendering over crashy
rendering ;)
- instability in the benchmark. Can you provide more details about how
you obtained these figures? Are they averages over multiple runs?

> table:
> it took 27 seconds to open the whole table, which is 10 seconds slower
> than on 2.14.4

Better than 10 minutes though! Still, I've identified some more
culprits causing slower attribute table opening - working on a fix
now.

On that note -- I've nearly exhausted the original allotted 3 days for
2.18 bug fixes. I've setup a good framework for identifying speed
regressions like this and I'd like to identify (if not fix) the cause
of these regressions (and possibly, if I have the time, also fix the
extreme performance regression in the mssql provider described in
https://issues.qgis.org/issues/15752 ). There's also still a
frequently occurring crash which occurs in composer after pasting
items which I'd love to fix before 2.18 is made LTS. Is there any
chance I could get funding for another 3 days of bug fixing?

Nyall