[Discussion] Preparing for the upcoming v0.19 release

[uwestoehr]

Yes, I have been looking into fixing Sketcher bugs and some PartDesign ones.

A big thank you! You fixed annoying and long-standing bugs.

For some of them I cannot reproduce (I have posted in the relevant threads). I think it would make sense to close those in which nobody can no longer reproduce.

Yes, please just close them. If the reporter can nevertheless reproduce, he will reply. This way of handling bugs is in my experience a good way. It might appear harsh but we did so in the Software project LyX (that I worked for before I came to FC) and it turned out to be effective since before, we smothered in bug reports and it cost us so much time to look through them again and again.

Do we have a release criteria? (a preliminary release date or a minimum number of features or fixes we want to include)

Good question That there is no such criteria, nor a release maintainer etc. is the last FC mystery for me

Werner, Yorik, can we please set up something transparent to get to a release? What about a Wiki page where you input the TODOs from your perspective? Others can have a look and help out in reviewing pending PRs etc. I mean we have some manpower, so some person can be the release maintainer for translations, another one for the documentation, another one for pending PRs, another one for the website (news etc.) and another one for the open bugs.
This way we distribute the workload and encourage the developers since we have a transparent way for a release and everybody can join one of the mentioned fields.
Can you both please make a statement accordingly?

Since there is nothing better than doing it, lets ask around who wants to take over a part for the release? I set up a Wiki page to fill in the names: Release_of_FC_0.19

EDIT: BTW, I have created the 0.20 version in the bugtracker,

I still really believe version 1.0 should follow 0.19 if realthunder's toponaming stuff went in.

.

[vocx]

...
Werner, Yorik, can we please set up something transparent to get to a release? What about a Wiki page where you input the TODOs from your perspective? ...

See this page, Release_process. It was created in 2014, but hasn't been used really.

People have good intentions, but often fail to follow through with concrete plans. Today there are more contributors so maybe now it's possible to get organized.

[chrisb]

See this page, Release_process. It was created in 2014, but hasn't been used really.

The strict time intervals are sort of daily trivia. The important thing is that Werner or Yorik have to announce a feature freeze and a release date. And the rest of us is doing whatever is possible. Everything else may be too much overhead.

[abdullah]

Good question That there is no such criteria, nor a release maintainer etc. is the last FC mystery for me

As Chris says, all starts with the feature freeze. I have not seen it announced yet.

In the past there have been good reasons to have longer term releases, for example PartDesign. My question was more in the sense of: Are we waiting for something specific? Is there any part of FC or lack of stability that is preventing a release?

The reason for the question was mainly to see if I somehow can help and also to organise my development (should I start a new feature now or should I continue bug fixing?).

I still really believe version 1.0 should follow 0.19 if realthunder's toponaming stuff went in.

It has been long discussed and I am not willing to unearth the cadaver. Whatever number it is, I needed a place to push tickets that will not be handled in this development cycle (such as new features, specially if we are close to a feature freeze).

[uwestoehr]

As Chris says, all starts with the feature freeze. I have not seen it announced yet.
... Are we waiting for something specific? Is there any part of FC or lack of stability that is preventing a release?

This is my question too. Therefore I set p the Wiki page that we can collect the musthaves: https://wiki.freecadweb.org/Release_of_FC_0.19
I hope now that people fill the page to be able to see what needs to be done.

I still really believe version 1.0 should follow 0.19 if realthunder's toponaming stuff went in.

I needed a place to push tickets that will not be handled in this development cycle

Sure. There is no need to wait with this. Just push bugs to the 0.20 target. The target number can anytime later be changed to 1.0 or whatever.

[yorik]

There has not been any feature freeze announcement yet (unless I missed it). Draft is not a problem, I use it everyday and there are no serious bugs at the moment. The main problem is the huge amount of realthunder's work in the pull requests queue.

We cannot simply merge everything at once, because we'd probably be in for huge periods of heavy instabilities, with only one person (realthunder) to fix these, and I'm sure everybody will agree it's not healthy at all for a project to have such a big dependency on one single person, even realthunder being very careful and listening to everybody and fast to solve problems.

Many of realthunder's features currently in master are still half-baked and many of his subsequent PRs are needed to fix everything. So IMHO it also doesn't make much sense to "leave as is". And we'd be only pushing a problem for later (which is also never a good idea).

To resume, as much as I would like to, I don't think we should release before having advanced a bit more in Realthuder's merges. Many of them are huge, with many changes all over FreeCAD, and pretty much impossible to assess without testing. Realthunder's opinion would be good to hear on that subject... are you around @realthunder?

I am myself not skilled enough with C++ to decide on Realthunder's code, and I believe nobody is (except Werner maybe). But any of us can help with testing. Werner doesn't do differently, AFAICS. If we test a PR and can write a review saying that we tested and report what we found, we can help greatly the process to go forward. If several of us (specially C++ people please) do test and approve the code, I think we could build enough trust to merge with far less need of Werner's work.

[chrisb]

Just notify realthunder:

ping

[triplus]

In addition please make an "official statement" about the next release being 0.20 or 1.0 oriented. Some of us have invested a few years of effort in this area. In discussions, aligning opinions, understanding the key points, gathering consensus on what all agree 1.0 should be about. This effort has matured and the arguments for doing a 1.0 are strong now. More or less all agree TopoNaming is the remaining big thing to do, and as that will be a part of the next development cycle, there is no real reason, on why the next development cycle shouldn't be 1.0 oriented.

P.S. Arguments against doing 1.0 and to wait a few more years, like improving documentation, more testing ... are in my opinion weak. Anyway, it's time now, to decide, regardless of the outcome, people can after move on.

[chrisb]

All arguments and options (0.20, 0.9x, 1.0) and links to former discussions are on the table: https://forum.freecadweb.org/viewtopic.php?f=8&t=43461.

[Zolko]

In addition please make an "official statement" [...] Some of us have invested a few years of effort in this area.[...] Anyway, it's time now, to decide

I think that's the most important issue to do to 1.0: what is the decision process at FreeCAD ? If the mantra continues to be "it's done when it's done" then that excludes 1.0 for me. Whatever the features. If, on the other hand, an official body exists, with transparent decision-making process, then the technical hurdles can be overcome.

I – and I think all professional engineers will feel the same – cannot invest a big project on a tool decided by unknown people on their whim. It's as simple as that, all the rest is poetry. It won't prevent FreeCAD to be a nice tool for hobby designers of course, but it will not compete with commercial products, even if though it's free.