Will Git change the FreeBSD development model?

[bobmc]

Why Linus release git instead participating to mercurial ?

Mecurial is written in Python. lt is probably slower than git and may have limits for large projects.

But Mecurial has a sophisticated model similiar to git.

 

[ralphbsz]

Yah, if you want to change the original then I fully agree. Unfortunately that's not the only restriction, if you want to use the original and have it become part (unchanged!) of your own project you're still forced to release all your stuff under the GPL license.

That's the part which irks me and which I simply cannot (and do not) respect.

That depends on what you mean by "use". If I write some code that runs in userspace on Linux, and is compiled with GCC, and uses glibc, it does not get infected with the GPL. I can release it commercially without releasing its source, I can release it under BSD/MIT/Apache/... license, or whatever I want. Similarly, if all the code does is link against libraries and call functions from libraries that are distributed under the LGPL, then my code does not get the viral infection.

On the other hand, if the code bodily includes GPL'ed source code (for example contains a copy of a device driver from the Linux kernel, or copies 100 lines for lock management from the linux kernel), then it is subject to the terms of the GPL. And these examples are not chosen arbitrarily: I've worked with people who thought it would be cool to copy 100 lines of lock management from a (non-open-source) operating system, and another person who wanted to use a device driver from Solaris (!) which they had obtained the source for illegally (we did not have a source license). The first person was forced to rewrite the locking code in their own way; the second one was fired (stealing source code while at work is not funny).

So I can understand that the authors of GPL'ed software put on restrictions. Now, do I like the specific restrictions of the GPL? No. They are not sensible or reasonable, they are an embodiment of RMS's crazy political ideas about all bits wanting to be free. He also has other political ideas, about sex with teenagers, which caused him to lose his affiliation with MIT. I personally think that RMS is nothing but a washed-up hacker (*) from the 70s, who took some not-thought-through ideas from the 60s and popularized them. Fortunately for him and unfortunately for the software world, he was at the right place at the right time for these ideas to get lots of traction. And given the permanent viral nature of the GPL, that's really hard to undo, once the software is in widespread use (which Linux is, in particular through being in a large fraction of all cellphones in the world, by virtue of Android). But freedom also means freedom for others, and the people who wrote Linux (and gcc and glibc and the Gnu utilities) made a choice.

(Footnote *: The word hacker here used in the positive sense.)

 

[fraxamo]

Came across this quote today by Warner Losh (a member of the FreeBSD Git Working Group) in an article on the FreeBSD Foundation blog:

<snip>
"Release 13 marks the first major release since the transition to Git. Not only does Git accelerate the onboarding of new contributors, says Losh, as a distributed version control system, it also facilitates collaboration with FreeBSD committers and non-committers alike."
</snip>

Here's hoping that the move to Git does indeed increase contributions.

 

[zirias@]

Sure, cause it should make things easier. Even as a non-committer, you can profit a lot from local branches and being able to rebase your stuff to form a submission for review that is a set of logical, self-contained commits.

Still, it won't change anything about the development model. For good reasons.

 

[Jose]

...Even as a non-committer, you can profit a lot from local branches and being able to rebase your stuff to form a submission for review that is a set of logical, self-contained commits...

Rebase-squash is so great for reviews.

 

[zirias@]

Rebase-squash is so great for reviews.

Hmmm. Squash "depends". For a complex change, a series of commits makes sense, but it should follow some logical path. I think git is a great tool for creating such things

 

[kpedersen]

I am quite a fan of Git. I think it is a little overkill for some projects but at the same time I use it for many personal projects (even tiny ones).

What surprises me more (this might sound perhaps a little unfair) is that it managed to come out of the Linux project!

Linux does like heavy things (and unless you use Git/SSH, the common approach is to use very heavy web technologies (like gitea, gitlab, github Enterprise) to host little repos) but the more agile distributed (non-centralized) approach is something that Linux (and GNU) is not great at.

So I don't think Git will really change FreeBSD's development because in some ways it is not very Linux-centric in the first place.

 

[bobmc]

Will Git change the FreeBSD development model?​

The development model is determined by policy (rules for developers), not mechanism (the VCS commands). Git (or Mecurial-scm, for example) can be used by a team with no central server but it is not. So it seems the same as using SVN and CVS. Some code changes are accepted with patches via email, instead of merging branches in the VCS. This is policy and I don't expect it to change.

 

[zirias@]

I am quite a fan of Git. I think it is a little overkill for some projects but at the same time I use it for many personal projects (even tiny ones).

Git has its pecularities, and there's a learning curve using it that could be more gentle. BUT: It's a distributed VCS that works well, and distributed gives you a lot of advantages (one of them being that you can start a project and track changes locally and postpone publishing the whole mess to whenever you think it's a good time). Among all the distributed VCS systems, git is by far the most popular. Maybe others are "better" depending on what you focus on. But git gets the job done and is "quasi-standard". Most developers will, by now, be familiar with git, so, it's the first candidate to consider when moving to a distributed VCS.

What surprises me more (this might sound perhaps a little unfair) is that it managed to come out of the Linux project!

No comment on THAT. (which kind of is a comment as well ?)