Assembly3 preview

[fosselius]

"let's make it simple. Then we can improve things from there. Some people would have to swallow their pride: tough, that's life"

This have nothing with pride to do, this is a big decision with big impact on the project, this will potentially cause additional work for a lot of developers in the near future. and might result in a worse user experience.

That said, I feel that the LinkStage3 branch is well tested, but i do not know when it was last rebased against master.

[HarryGeier]

Hi,

i am following this since it started. I am no developer myself.( this would surely mean a divorce and early retirement for me ) but i am a good programmer.

Now it´s a point in time where both , the core freecad dev´s and realthunder must decide where to go from here. I only see 2 possible scenarios :

1 : Assembly3 ´s core deviations are integrated and make it possible that the extensions introduced in the fork are getting possible to be done one after the other and the fork is freezed. ALL concentrate on the merging in an early 0.19 stage.

2: Assembly3 is taken out of of the whole FreeCAD world and becomes it´s own project, logically with it´s own name and forum etc.

We cannot afford that developers , testers and users a re constantlly talking about FreeCAD but not meaning the same. All development must concentrate on bringing ONE FreeCAD closer to perfection. I know some will say that a split would mean the end to assembly3 , because a single developer , as big a genious he may be, can´t raise and maintain a mainstream opensource project at that size. But it was realthunder´s decision to go as far as he is away from the standard. Surely, FreeCAD´s main development will be influenced by the merging an probably create much hassle to , but i really see no third way.

Quo Vadis ?

[wsteffe]

I only see 2 possible scenarios :

I also think that it is time to decide: or assembly3 is accepted for merging or it becomes a new project. And I would like to see that the first choice taken.

Regarding the way of merging, I agree with Fosselius and Zolko and I would consider a waste of time splitting the fork in many parts and merging them one by one until the Linkstage3 fork is recreated (as result of this stupid process).
If Linkstage3 is well tested and works well, as reported by many, than the merge should be very easy. Linkstage3 could simply become the new master.

But, according to what said by Realthunder, Linkstage3 should be a couple of months behind Master. Therefore I think that, instead of freezing the fork as suggested by HerryGeier, it would make more sense to freeze the Master until the developers have realigned the Linkstage3 with Master.

[saso]

If Linkstage3 is well tested and works well, as reported by many, than the merge should be very easy.

Really? Who, besides realthunder (I guess he does) actually understand all of the different options and modes that are available in his fork?

[wsteffe]

Really?

I am not sure of that. In fact I used "If" in my statement.
More tests perhaps should be done to verify that Linkstage3 doesn't introduce substantial regressions.

all of the different options and modes that are available

To me it would be sufficient to acknowledge that Linkstage3 does what master does (without any regression) besides other new things.
If there are some unknown/untested options I do not care too much.
We may see in the future if those options are useful or not and it will be possible to remove them if not.

[saso]

To me it would be sufficient to acknowledge that Linkstage3 does what master does (without any regression) besides other new things.
If there are some unknown/untested options I do not care too much.
We may see in the future if those options are useful or not and it will be possible to remove them if not.

If it would be just about comparing it to master and testing for regressions I would agree, my feeling and why I have doubts about merging all of it at once is however that what we individually understand and find as useful in this fork could very well be just 10 or 20 % of all the things that have been developed there. With this I am not saying that only 10 or 20 % of it is good and useful, just that there is so much that I doubt that anyone has a good overview and understanding of all of it. And there are for sure things that are good and could be merged now, but there are also things that should probably be removed or reworked...

[Jee-Bee]

The main issue is that the link stage branch is according to now:

This branch is 1229 commits ahead, 867 commits behind FreeCAD:master.

This is a huge amount of differences. I can remember that to handle the changes in master related to his own branch.
If @realthunder have difficulties with it. in 0.16 part design (next) is added to master and it takes almost 2 years!!

It was better if the discussion how to handle changes has started earlier...

[wsteffe]

there is so much that I doubt that anyone has a good overview and understanding of all of it

An injection of many code lines that nobody (apart of RT) understand could actually lead to troubles if it obfuscate the FC code
to other developers. But instead of removing the undocumented code perhaps this problem could be better addressed asking for a
more detailed documentation to Realthunder.

[Zolko]

It was better if the discussion how to handle changes has started earlier

He (realthunder) tried but the "core developers" actively ignored him and his proposals.

Let's face it: what is more difficult now, taking ovet LinkStage3 as-is and trying to understand it, or redeveloping the same functionality from scratch ?

[dulouie]

LinkStage3 documentation is here:
https://github.com/realthunder/FreeCAD_ ... /wiki/Link
https://github.com/realthunder/FreeCAD_ ... re-Changes

And i'm sure realthunder will do rebasing linkstage3 branch, to get even with master.