[Qgis-psc] Performance regressions (was QGIS 2.18 bug fixing effort)

[Giovanni Manghi]

Hi Nyall,


> - it's not possible to directly compare the repo provided versions vs
> self-compiled builds. They'll be built using different compilers with
> different optimisation levels. Are you able to re-run the tests which
> used the repo builds with self-compiled builds instead?

yes I can, and will do tomorrow morning.


Also please
> make sure you're building using "Release" type builds, not "Debug".
> (CMAKE_BUILD_TYPE should be "Release").

this was already set the right way in my previous tests.



> - Can you share the script you used to record the benchmarks?
> Specifically, I'm wondering exactly what is being timed here. Is it
> using the command line options to render a map (and accordingly
> including project and layer load times), or timing canvas refreshes,
> or using some custom Python script?


no script, I'm just using the "time" command to launch qgis from the
CLI and open my large testing dataset (so measure how long it takes to
fully load from the moment is added in QGIS). From the output
measurements the one that should be taken into account is "user" I
guess.

Times to open attributes tables were taken not in a precise way, but I
don't think that here a few (few) seconds make a lot of difference.
Anyway I'm still a great supporter of Alessandro PR that would allow
to load the tables in a different thread.



cheers!

-- G --