[RFC] a 0.19 release right now with Py2 and Qt 4 support?

[uwestoehr]

I joined FC a year ago and I am impressed by the improvements of FC during that period.
Last week I recommended a colleague FC to design parts to be 3D-printed. It was a success since she tries now out FC for that task. Of course she initially went to our website and downloaded the latest stable version, which is 0.18.4. I recommended her to use FC 0.19 and showed her some nice improvement of e.g. the meshing. Thus she asked me why there is then no more recent stable FC release.
Good question in my opinion.

I already thought of proposing a release soon because I think we could do it this way:
- release soon to enable user to benefit from the many improvements
- officially support Python 2 and Qt 4 for this release the last time and drop this for future releases
- freeze the development of new features for e.g. a month to get time to fix bugs and update the docs. Especially the latter will need time.

What do you think?

[vocx]

...
What do you think?

This is discussed here: [Discussion] Preparing for the upcoming v0.19 release

Summary:
* Werner doesn't want to rush the release, he wants to solve some pending issues before that.
* We were thinking about freezing it by mid-December 2019, in order to test the software enough, so the new version will be included in the next Ubuntu LTS (April), which requires a stable version around February.
* Since nothing has been officially announced, I assume that at the current pace, the official release will come around March or April.

[uwestoehr]

* Werner doesn't want to rush the release, he wants to solve some pending issues before that.

I fully agree with him in this respect because I see growing number of confirmed bugs. The point is that we are still input new features and they require out time. For example, every new preferences option in a WB needs to be documented and tested.
Therefore a feature freeze would help the project in my opinion to stabilize the different WBs.
And a release is the best reason for a short freeze

I understand the conflict with a freeze since adding a new feature is more fun than fixing bugs. For example the last days I tried to fix some TechDraw bugs but could not resist to make small improvements too.

This is discussed here: [Discussion] Preparing for the upcoming v0.19 release

Thanks. I only searched the developer subforum before I posted.

[sgrogan]

Maybe it would be good to look here: https://forum.freecadweb.org/viewtopic. ... 40#p356708
And think about 0.18.5.
I'm not advocating an onslaught of back-ports, but some might make sense?

[looo]

* Werner doesn't want to rush the release, he wants to solve some pending issues before that.

Py2 support is becoming more difficult. I have decided to stop supporting py2 with external workbenches. I hope for a soon release and afterwards we stop python2 support.

[apeltauer]

I agree with Werner not to rush the releases. But what about creating a daily/weekly build and upload it to the website? OpenShot is doing this too. see https://www.openshot.org/de/download/ it is not actual a daily build but i think some of the latest code states. I know we have conda, but i think this is not a handy way for users who are not familiar with conda (so i am ).
So what about creating a automatic build via travis and upload it fully automatically once a week/day?

[wmayer]

* Werner doesn't want to rush the release, he wants to solve some pending issues before that.

Py2 support is becoming more difficult. I have decided to stop supporting py2 with external workbenches. I hope for a soon release and afterwards we stop python2 support.

After the release Py2 and Qt4 support will be definitely stopped. For me the only reason to still support Py2 is that we are in the lucky situation that no major change was made so far that would completely break Py2.

I already thought of proposing a release soon because I think we could do it this way:
- release soon to enable user to benefit from the many improvements
- officially support Python 2 and Qt 4 for this release the last time and drop this for future releases
- freeze the development of new features for e.g. a month to get time to fix bugs and update the docs. Especially the latter will need time.

The earliest date of a feature freeze might be mid/end of February because there are still some serious issues to fix and I also want to merge most of the pending PRs.
This time I didn't want to spend my whole Christmas vacation for code reviews or bug fixing like in the recent years and the last two weeks I wasn't at home and thus couldn't do much.

[looo]

So what about creating a automatic build via travis and upload it fully automatically once a week/day?

We are doing this for quite a while. Windows is missing because we haven't found an alternative appimage/dmg for windows. Also we enjoy packaging support from @sgrogan. If you know how to create a portable for Windows, it shouldn't be too difficult to add it:
https://github.com/FreeCAD/FreeCAD-AppI ... ster/conda

But supporting different python versions with weekly builds is not feasible (imo) for sure anybody can step in and do the work to support py2. But py3.8 ist now showing already what is meant by "python2 support has stopped in 2020".

[Zolko]

Py2 support is becoming more difficult. I have decided to stop supporting py2 with external workbenches. I hope for a soon release and afterwards we stop python2 support.

After the release Py2 and Qt4 support will be definitely stopped. For me the only reason to still support Py2 is that we are in the lucky situation that no major change was made so far that would completely break Py2.

What is the benefit of supporting Py2/Qt4 ? If the overhead is small but non-zero, and since developer time is scarce, and since v0.19 is a major upgrade, I think that the ratio cost/benefit would be better for the general FreeCAD community to completely remove any Py2/Qt4 support from the 0.19 release and use your valuable time for other purposes.

[wmayer]

What is the benefit of supporting Py2/Qt4 ?

Because from what you can read in the internet still many people use Py2 because they can't (or don't want to) switch to Py3 due to dependencies to other packages that are not ported. And Py2 support was only done under the premise that working time is low.

But you're right that for FreeCAD itself there is not much benefit.