[QGIS-Developer] QGIS Server and the Grants programme

Tue Jun 9 01:38:08 PDT 2020

Hi Andreas, (& All),
A fair point, but I believe this is an important point and this year I 
do have data to back up my point; in fact the grant program is what 
motivated me to finally get around to doing this analysis.

It seems from the replies that while there are a few differentiators, 
the key one is indeed cartography and styling. (There's also an 
interesting conversation about vectors going on there too). Some thoughts:
* The vast majority of WMS/WMTS layers are not cartographically 
complicated, let alone beautiful. They're "here is a layer with small 
green points for trees", and "this polygon represents conservation 
areas". You can easily play around and see what's out there here: 
http://www.geoseer.net/api-demo/
* WFS/WCS can't be styled server side.
* It seems like overkill to create and maintain an entire server 
distribution that fundamentally only solves one (relatively simple 
compared to what QGIS Desktop can do) problem.
* Rendering is only one part the QGIS package (Analysis, digitisation, 
management, etc.).

If I'm honest, the "competition" on this point isn't really between QGIS 
and MapServer/GeoServer. It's really between QGIS and ArcGIS. Because 
ArcGIS does exactly what QGIS Server seeks to do: offer a single 
integrated solution for Desktop-> Server. And certainly ArcGIS Server 
does have a huge number of deployments (53%), however again, there 
really aren't many cartographically complicated outputs on there. And 
despite the huge number of deployments, most services and datasets are 
actually served by MapServer/GeoServer (about 60% of datasets between 
them!). Basically ArcGIS is deployed by local government and used for 
bitty-stuff ("here are our fire stations"), but if you want a real 
data-service then you go with GeoServer/MapServer/etc.

Most importantly though, I think I haven't conveyed my core point well: 
this really is a zero sum game!
Even allowing for the above, any funds spent on QGIS Server are not 
spent on QGIS Desktop. There are 60 public facing QGIS Server 
deployments. Even assuming that there's a ratio of 10:1 for 
private/public servers (made up ratio, feels too high), any funding on 
QGIS Server benefits only hundreds, or being very generous, maybe 
low-thousands number of users. Funding on QGIS Desktop however benefits 
as a *minimum* tens of thousands, potentially millions of users (no idea 
how many QGIS installs there are, I can't find the download-stats I 
remember seeing in the past).
Heck, even pretending for a second QGIS Server had 100% of the 
deployments (a 100 fold increase!), there would /still/ be orders of 
magnitude more people using the not-Server parts of QGIS Desktop by its 
very nature.

There are 3,102 open issues on the QGIS issue tracker. 95 are labelled 
regressions, 137 are "high priority", and 92 are "crash/data 
corruption". Just 49 are "Server". I'm not seeking to denigrate the 
project here; QGIS is a extremely complex tool that is an amazing 
accomplishment and by its nature it will have bugs. I raise these 
numbers to highlight that any money spent on Grants to Server (and yes 
new Desktop features) is money that isn't spent fixing these (I'm aware 
of the bug-fixing fund). Something I think the grant voters should be 
cognizant of.

Hope that clarifies,
I'll step back now. :-)
Cheers,
Jonathan

On 09/06/2020 08:09, Andreas Neumann wrote:
> Hi Jonathan,
> You keep repeating yourself. You started the exact same discussion a
> year ago.
> You have a valid point, of course, I don't argue that. But if you think
> about small organizations  that do not have a lot of personal (or
> financial) resources, it would be a lot of burden to invest twice the
> time in styling: once for QGIS desktop and another time again for UMN
> mapserver and Geoserver. Even if SLD output from QGIS improved (also
> thanks to efforts of Andrea Aime and others), it still can't transport
> everything. If it would, then I would better agree with your argument.
> For such smaller organization, speed (and I know that UMN and Geoserver
> are a bit faster than QGIS server) is not the only important thing - it
> is also their personal and financial resources and complexity of their
> software landscape.
> And QGIS server has some other unique selling points: the proprietary
> GetPrint command that doesn't have a match in Geoserver or UMN, the
> ability to create Atlases from server, and who knows, in the future
> perhaps QGIS server can run processing models.
> Greetings,
> Andreas
> On 2020-06-08 22:42, Jonathan Moules wrote:
>
>> Hi List,
>> Some of you may have seen my blog post on the OSGeo-Discuss list 
>> about which mapping servers are the most deployed. For those who 
>> haven't seen it, QGIS Server has about 60 public deployments (1% of 
>> all of them), and it serves 11,924 datasets (0.5% of all public 
>> geospatial WMS/WFS/WCS/WMTS datasets).
>>
>> Potentially controversial here and I appreciate it's not a 
>> competition, but given the low uptake of QGIS Server compared to 
>> other Open Source offerings (GeoServer: 964 deployments, 963,603 
>> datasets; MapServer: 544 deployments, 389,709 datasets), is QGIS 
>> Server something the grant program should be funding? There are three 
>> Server proposals totalling EUR10,000, 22% of the fund.
>>
>> Now, before you get the pitchforks out(!), please consider the 
>> following:
>>
>> * Zero sum game - Any money spent on QGIS Server cannot be spent on 
>> QGIS Desktop. (The grants mostly aren't things that will improve the 
>> shared QGIS Core). (This reasoning also follows through to OSGeo funds).
>>
>> * Multiple solutions - Open Source (and OSGeo) already has a very 
>> healthy ecosystem of mapping servers - does it need another one?
>>
>> * Limited number of users benefited - I don't have stats for it, but 
>> QGIS Desktop is probably the most popular Open Source Desktop GIS, 
>> and is certainly going to have many orders of magnitude more users 
>> than QGIS Server.
>>
>> * Playing to your strengths - QGIS' strength is it's Desktop and it's 
>> generally good practice to play to your strengths.
>>
>> So given the above, and that QGIS is already "winning" as an Open 
>> Source Desktop (great job!), I'd like to suggest it's not a good idea 
>> to dilute the limited resources by spending them on QGIS Server. 
>> Instead it seems that far more people would benefit if that money was 
>> spent on Desktop, especially the bug fixing programme.
>>
>> Or alternatively, given the "Unique Selling Point" of QGIS Server is 
>> its integration with QGIS Desktop, those resources could be used to 
>> further improve interoperability with 
>> GeoServer/MapServer/deegree/etc. Those are all successful mature 
>> OSGeo projects that excel at serving maps, have an architecture 
>> designed for it, and already have huge install bases.
>>
>> TLDR: QGIS excels at being a Desktop, and I'd like to suggest it 
>> should play to its strengths and focus its limited funds there to 
>> benefit the most users.
>>
>> I shall now retreat to my bunker. :-)
>>
>> Cheers,
>> Jonathan
>>
>> Note: The above only applies to the Grant program and funding; how 
>> developers wish to spend their time, and on which projects is of 
>> course their own prerogative.
>>
>> (Disclosure: I have no horse in this race; I don't run or administer 
>> any mapping servers, but I have done GeoServer in the past.)
>>
>> _______________________________________________
>> QGIS-Developer mailing list
>> QGIS-Developer at lists.osgeo.org
>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20200609/6c328f17/attachment.html>