[QGIS-Developer] QGIS 4.0 - let's start some early discussions!

Nyall Dawson nyall.dawson at gmail.com
Tue Jun 25 20:28:40 PDT 2019


On Tue, 25 Jun 2019 at 21:13, Martin Dobias <wonder.sk at gmail.com> wrote:

> Yeah it would make sense to align QGIS 4 with a jump in technology
> (i.e. Qt5 -> Qt6). Maybe we could also investigate the option to use
> Qt for Python (which is now a fully supported part of Qt) and free us
> from PyQt and SIP (but the question still to be answered is whether Qt
> for Python is at least as good as PyQt implementation).

This seems like a recurring theme. Let's add it to our api break
tracker so we've got a record.

For me, the biggest pain point in current API relates to all the use
of CRS database id values. These are redundant, fragile, and block us
from some nice features we should have through proj 6. I'd very much
like to kill these and require AUTH:ID codes everywhere, but that's a
break. I've hacked around it as best I can for now, but it IS limiting
the extent we can utilise the new world of projections.

> For timing how about something like "not earlier than in 2 years and
> not later than in 4 years" :-)

Works for me. Especially if Qt 6 is targeted for end of next year,
that would give us ~1 year to hold off a transition. (Although,
potentially if Qt6 is a minor break we'll likely be able to get qgis
building on both 5/6. It's just... plugins... which are the issue.
again.).

Nyall

> My observation is that lots of people are still using 2.x as we kind
> of made people scared of upgrading (due to the API changes, missing
> plugins etc.) and they got quite conservative. As people only now
> slowly move to 3.x I think that's another reason why we should not
> push for a 4.0 release too early.
>
> In any case, this is a very useful discussion to have - and get
> everyone on the same page.
>
> Cheers
> Martin


More information about the QGIS-Developer mailing list