[Qgis-psc] Funding QGIS Server enhancement - remove old QgsWXSProjectParser

Wed May 24 08:29:37 PDT 2017

Hi all,
we had a chat with 3liz to share a common view of current refactoring
status and next priorities:

*What the refactor is expected to adress  :*

 It is a blocking part since it is quite hard to find funds for "plumbery"
tasks and make it understandable by users.
 The refactor is obviously made mandatory by the move to QGIS3 API.
 Obviously again, it will help in having a clearer codebase, easier to
maintain , faster and more modular.
 For instance, having the ability to generate on the fly some project and
layer definition will help a lot in GeoNode integration.

*What has been done :*

 - *3liz* has created a new *modular architecture* allowing to refactor
each services independently. It exposes a clean API, and if I get it well,
*Alessandro* already used it to rework the server plugin python API.

 - *3liz* is refactoring progressively all the *services*. All WFS, all WCS
and WMS GetCapabilities are done. GetContext is being worked on.

* - Oslandia : *Paul is almost done with a full WMS rewrite now only using
core classes and no more XML parsing

 - *Oslandia : *OGC CITE compliancy docker testing framework - see [1]

 - *Oslandia* : Speed improvements for getcapabilities where datasource are
massive views or materialized views (or layers with definition filters /
joins). AKA "trust project" option.

 - *Camptocamp* created a docker platform for performance benchmark between
qgis server / mapserver and geoserver - see [3]

*What needs to be finished before 3.0, where funding will help a lot :*

 - *all*: a lot of work has been done to remove the projectParser and
singletons, but there is still some work to do.

 - *services rewrite* *(3liz)*:  getContext / DescribeLayer /
GetFeatureInfo, GetStyle, DXFWriter

 - *service rewrite (Oslandia): *GetLegendGraphics / GetPrint

*What would be (really) nice to have :*

 - *OGC CITE (Oslandia)* : Automate running and publishing html reports to
a public ressource, if possible hosted by QGIS.org. (small task)
 - *OGC CITE* : debug existing failures and apply for an official OGC
certification. No idea how much time this can take.
 -* Performance benchmark platform* : define reference tests, automate them
and track regressions or improvements over time.  This is a tricky topic,
but performances tests with reference platforms helped a lot in raising
some issues.
 - *Communication* : push a blog post from QGIS.org for an official support
of the project and advertising there all the funders and involved parties.
That would help a lot in raising new funds too..

*What is not yet planned :*

- Implement a caching strategy for scalability
- vector tiles ?

note: A detailed view is available in QEP74 [0] and all related PR are here
[4].

So, in short, if some funds are available from QGIS.org, I think we have
some candidates and priorities to discuss here. We also need to estimate
more precisely each task...

Opinions welcome !

[0] https://github.com/qgis/QGIS-Enhancement-Proposals/issues/74
[1] https://hub.docker.com/r/oslandia/qgis-server-ogc-cite/
[3] https://github.com/camptocamp/ms_perfs
[4]
https://github.com/qgis/QGIS/pulls?q=is%3Apr+label%3AServer+milestone%3A%22QGIS+3%22+is%3Aclosed

2017-05-21 16:35 GMT+02:00 Tim Sutton <tim at qgis.org>:

> Hi All
>
> From my side I am ok with QGIS.ORG funding work where we feel it is in
> the good of the project. I think the simple solution if we have the cash is
> to just raise a Loomio proposal with the Voting Members and let them
> decide. I think there should be some leeway for QGIS to make discretionary
> sponsorships of features or bug fixing etc. outside of the normal budget
> plans (so that we can e.g. capitalise on budget savings as per your example
> above) but I am worried that we will appear to subvert the QGIS Grant
> Programme process if we do it in this case. Getting it done by a loomio
> vote (with a nice clear explanation of why it is useful) should be quick
> and easy and  transparent.
>
> BTW I am also very -1 on you spending your personal cash on this - we (
> QGIS.ORG) already use a lot of your time (which equates to lost earnings
> potential) via your work in the PSC and I would not like to see you putting
> in your cash too when probably the community would be happy to fund it.
>
> I am also thinking it might be useful to broaden our scope a bit and
> collect up any other QGIS Server work that needs funding into one work
> package and make the proposal to fund everything that we can. I would not
> like to see QGIS Server get left behind as we forge on to QGIS 3.0 and we
> should support it as strongly as possible from the QGIS.ORG funds. I
> think this approach might also be more effective than a crowd funding drive
> too which may result in a lengthy wait for no result.
>
> Regards
>
> Tim
>
> On Sat, May 20, 2017 at 9:37 PM, Régis Haubourg <regis.haubourg at gmail.com>
> wrote:
>
>> Hi All,
>> Sorry for being late here.
>>
>> The big picture of what needs to be done has been drawn at the codesprint
>> in Lyon [0].
>>
>> We know that Camp2Camp has a benchmark framework that would address the
>> performance driven coding.
>>
>> 3liz is involved in adding more modularity and refactors other services.
>>
>> Alessandro Passoti is committed in refactoring the plugin API.
>>
>> Beyond those cleanups or rewrites, we also identified the need to find a
>> common strategy to allow scalability of QGIS server (caching layers and
>> project over server clusters)
>>
>> At Oslandia, we have funders that helped us in refactoring of WMS
>> services, OGC compliancy, security and performance issues. We keep working
>> on these items on our owns resources.
>>
>> Currently, our partners claim for some feedback before considering
>> funding more work :
>>
>>  - Having some official feedback that this work is appreciated and
>> welcome by the qgis.org project
>>
>>  - Having some references (web pages, blog, talks) showing that pushing
>> this work is triggering commitment from others parties, if possible other
>> corporations.
>>
>>  - and of course, see some real preliminary results of the already done
>> tasks.
>>
>> I think that the PSC could help in the communication plan to help funders
>> feel that they are not alone, and that there is a large community trusting
>> in the future of QGIS server and working hard on it.
>>
>> On our side, we have some web articles ready to publish, we will have a
>> common talk with 3liz at the Foss4G Europe in Paris.
>>
>> That said, any additional funding is more than welcome, but I'd prefer to
>> set up a common strategy than individual fundings.
>> Crowdfunding initiative in France are almost impossible to raise for
>> private structures, so if QGIS.org is going this way, we'll need a proxy in
>> front the different involved parties, and we'll have to coordinate on the
>> funding attributions.
>> Last point, Nyall has already a crowdfuding campaign opened, so not sure
>> if that could interfer or not (and we need that composer rewrite :) )
>>
>> Cheers, and thanks for all the good energy being given here!
>>
>> [0] https://github.com/qgis/QGIS/wiki/Code-Sprint-QGIS-3-Lyon,-end-2016
>>
>> 2017-05-20 15:10 GMT+02:00 Matthias Kuhn <matthias at opengis.ch>:
>>
>>> On 5/19/17 5:15 PM, Paolo Cavallini wrote:
>>>
>>> > Il 19/05/2017 17:06, Matthias Kuhn ha scritto:
>>> >
>>> >> A very good idea indeed.
>>> >> Are you proposing to have someone private take the lead, are you
>>> >> thinking of 3liz as developers or are you speaking of "we" with the
>>> >> Faunalia hat? Anything will do, just to make sure nobody feels
>>> >> responsible :)
>>> > sorry, I as in fact unclear: when I write here I speak as PSC member,
>>> so
>>> > we==QGIS.ORG PSC
>>> Sorry, I thought that qgis.org does not directly involve in crowdfunding
>>> projects, I must have missed the decision when this was changed. Are the
>>> requirements for a project to be executed by qgis.org available
>>> somewhere?
>>>
>>> All the best
>>> Matthias
>>> _______________________________________________
>>> Qgis-psc mailing list
>>> Qgis-psc at lists.osgeo.org
>>> https://lists.osgeo.org/mailman/listinfo/qgis-psc
>>>
>>
>>
>> _______________________________________________
>> Qgis-psc mailing list
>> Qgis-psc at lists.osgeo.org
>> https://lists.osgeo.org/mailman/listinfo/qgis-psc
>>
>
>
>
> --
>
>
>
> *Tim Sutton*
> QGIS Project Steering Committee Chair
> tim at qgis.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-psc/attachments/20170524/e8c7e0d3/attachment.html>