[Qgis-psc] PSC Meeting log from 8/4?

Vincent Picavet - ML vincent.ml at oslandia.com
Mon Apr 13 01:10:24 PDT 2015


Hi all,

I strongly support Nathan on all the points he mentionned.
We should clearly have a rough estimate of the next versions and the break 
they can introduce. 
Changes that require a new version would indeed be :
* Changes to the C++ API itself
* Removing old stuff
* Switch to Python 3
* Switch to QT5 and PyQT5
* Changing the Plugin model (a QGISPlugin class ?)

For our users, breaking the plugin compatibility has more consequences than 
even modifying the C++ API. It should not happen too often (no more than every 
1.5/2 years).

Another thing to take into account is the availability of QT5 and Python3 in 
the most frequent platforms and linux distributions. While we should be safe 
here, double checking would'nt hurt, or we will have to deal with a lot of 
backporting.

As for me, all these changes have to be in version 3.0, and as Nathan says, we 
need a good plan for this.

Having multiple big breaks within the next two years is not a good option 
IMHO.

Vincent

Le lundi 13 avril 2015, 00:29:38 Nathan Woodrow a écrit :
> Hey all,
> 
> I'm not sure I am super keen to make a 3.0 jump half way (or more) though a
> cycle.  If we are going to do 3.0 I think it needs to be planned out more
> and QEPs done for the major changes.  Not doing this risks leaving users in
> the dark on the API breaks and that never turns out well.    The main QEPs
> and changes that need to be reviewed for the changes are the new geometry
> stuff, PyQt5/Python3, removing old V2 methods and labeling stuff.  Might as
> well do it with a bang if we are going to do it.
> 
> If we wanted to go with 3.0 then I would suggest skip 2.10 and extend this
> period by 2 cycles so we have longer, get in review the API and QEPs, and
> have a plan rather then "just break whatever whenever and then release like
> normal".
> 
> I can understand that moving to PyQt5 might be a forced hand because of
> platform changes and that it is really going to break our API whenever they
> do it. If that is the case I think we really need a strong plan around it
> so we can have it all done at the same time.  No point in releasing 3.0
> with no PyQt5 and then having to break again for PyQt5 a few months later.
> Also mentioning that PyQt5 seems to be Python3 only at the moment which can
> also bring other issue so those need to be addressed.
> 
> As PyQt5 will break all the plugins again I think there needs to be some
> good education around it.    Python 3 has been out for years and people
> still don't move from 2.7, they didn't educate well and people didn't move
> then same can happen to us.
> 
> - Nathan
> 
> On Mon, 13 Apr 2015 at 05:35 Anita Graser <anitagraser at gmx.at> wrote:
> > Hi Nyall,
> > 
> > I've posted my log - bit of the beginning missing -
> > http://hub.qgis.org/wiki/quantum-gis/PSC_Meeting_8_April_2015
> > 
> > Here's the part on QGIS 3:
> > 
> > (3:51:39 PM) mhugent: So the question is how and when go for version 3
> > (3:51:58 PM) mhugent: E.g. QGIS3 after 2.10?
> > (3:52:16 PM) pcav: IMHO, since we have a LTR
> > (3:52:20 PM) pcav: we can go anytime
> > (3:52:25 PM) pcav: preferably soon
> > (3:52:38 PM) pcav: so people will have more safe time with the LTR
> > (3:52:42 PM) aneumann: Would 3.1 be another LTR then?
> > (3:52:52 PM) timlinux: yeah I also dont see much point in waiting till
> > after a 2.10 (3:53:12 PM) anitagraser: any time that works for you
> > (3:53:16 PM) timlinux: aneumann: yes
> > (3:53:24 PM) timlinux: 3.2
> > (3:53:26 PM) pcav: conservative peple can stay on 2.8 for 1 yr
> > (3:53:45 PM) pcav: and move to 3.2 after that
> > (3:53:48 PM) aneumann: it would be good to have something stable in Feb
> > 2016 (3:54:15 PM) pcav: right, that's why I'm suggesting to move soon
> > (3:54:23 PM) pcav: to 3
> > (3:54:42 PM) mhugent: For me it is also the sooner the better.
> > (3:55:12 PM) mhugent: jef, what is your opinion (since you are the release
> > manager)? (3:55:26 PM) aneumann: the question is if the ideas for the new
> > API are already ready/good enought to start with QGIS3? (3:56:15 PM)
> > aneumann: or maybe there are things to discuss before? (3:56:47 PM)
> > timlinux_ [~timlinux at 105-208-59-252.access.mtnbusiness.co.za] entered the
> > room. (3:56:57 PM) timlinux_: sorry folks I have to board my flight
> > (3:57:24 PM) jef: mhugent: I don't have strong opinions about version
> > numbers and api stability.   the api should never be turned inside out
> > and constantly changes anyway. (3:57:38 PM) timlinux left the room (quit:
> > Ping timeout: 246 seconds). (3:57:38 PM) timlinux_: +1 from me to make
> > the next release a jump to 3.0 but we need to hear from nyall (3:57:52
> > PM) aneumann: Nyall told me that he can implement things in parallel for
> > the print composer - similar to symbology 1/2 or labeling 1/2 (3:57:53
> > PM) timlinux_: because he may have specific timelines
> > (3:57:53 PM) pcav: +1
> > (3:57:59 PM) timlinux_: ok
> > (3:58:01 PM) anitagraser: +1
> > (3:58:09 PM) timlinux_: ok folks I will catch up in the logs
> > (3:58:16 PM) timlinux_ left the room (quit: Remote host closed the
> > connection). (3:58:28 PM) dassau: +1
> > (3:58:30 PM) aneumann: but that was because he did not get any feedback on
> > when QGIS3 will be started (3:58:45 PM) aneumann: Maybe he would prefer
> > otherwise
> > 
> > 
> > Best wishes,
> > Anita
> > 
> > 
> > On Sun, Apr 12, 2015 at 10:42 PM, Marco Hugentobler <
> > 
> > marco.hugentobler at sourcepole.ch> wrote:
> >> Hi Nyall
> >> 
> >>  I'm especially curious what "QGIS 3 (MH) [accepted]" means...
> >> 
> >> There was no formal voting, however all meeting participants shared the
> >> opinion that it is better to go for version 3 as soon as possible. That
> >> means there will be no 2.10 and the next version is 3 (API breaks
> >> allowed).
> >> 
> >> Regards,
> >> Marco
> >> 
> >> Am 11.04.2015 um 11:01 schrieb Nyall Dawson:
> >>> Hi PSC,
> >>> 
> >>> Was a meeting held on the 8th of this month? If so, is anyone able to
> >>> upload a log of it to
> >>> http://hub.qgis.org/wiki/quantum-gis/PSC_Meeting_8_April_2015 ?
> >>> 
> >>> I'm especially curious what "QGIS 3 (MH) [accepted]" means...
> >>> 
> >>> Cheers,
> >>> Nyall
> >>> _______________________________________________
> >>> Qgis-psc mailing list
> >>> Qgis-psc at lists.osgeo.org
> >>> http://lists.osgeo.org/mailman/listinfo/qgis-psc
> >> 
> >> --
> >> Dr. Marco Hugentobler
> >> Sourcepole -  Linux & Open Source Solutions
> >> Weberstrasse 5, CH-8004 Zürich, Switzerland
> >> marco.hugentobler at sourcepole.ch http://www.sourcepole.ch
> >> Technical Advisor QGIS Project Steering Committee
> >> 
> >> 
> >> _______________________________________________
> >> Qgis-psc mailing list
> >> Qgis-psc at lists.osgeo.org
> >> http://lists.osgeo.org/mailman/listinfo/qgis-psc
> > 
> > _______________________________________________
> > Qgis-psc mailing list
> > Qgis-psc at lists.osgeo.org
> > http://lists.osgeo.org/mailman/listinfo/qgis-psc

-- 
Vincent Picavet - Gérant Oslandia
www.oslandia.com



More information about the Qgis-psc mailing list