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

[Matthias Kuhn]

On 5/24/17 2:56 PM, ElPaso wrote:

> Il 24/05/2017 14:29, Paolo Cavallini ha scritto:
>> Il 24/05/2017 14:25, Matthias Kuhn ha scritto:
>>> I think this is still usable (qgis_bench IIRC).
>>>
>>> What it does: give usable measurements when executed on the same
>>> environment for different projects, different QGIS versions, different
>>> build parameters, different databases...
>>>
>>> What it not does: produce values that are comparable when executed on a
>>> non-stable system like travis CI. Performance on this system varies
>>> from
>>> day to day and we are never sure if a code change or some other change
>>> on the system (hardware, software, operating system...) is responsible
>>> for some measured "performance regression".
>> Thanks Matthias for reminding.
>> So it would be best to run it in an isolated, predictable environment.
>> Would a VM do? I think we could find the resources for it, even if this
>> would require having a dedicated server.
>> All the best.
>
> Maybe a Vagrant configuration to be run on AWS (or any other Vagrant
> compatible provider).
>
> What we probably need is something we can spin up and switch off just
> for the time required for the test, that's why AWS would fit.
There are some notes by Radim definitely worth reading (!!) here:

https://github.com/qgis/QGIS/tree/master/tests/bench


I think shared environments (most VMs / AWS? / cloud solutions) are
pointless.

A dedicated server might work but I'd consider even that an adventurous
project (i.e. worth trying but with no guarantee of success ;) )

Cheers

>
>
> Just my 2 cents.
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-psc/attachments/20170524/3f44aa8e/attachment.html>