Squeak - Dev mailing list wrote
> An enterprising dev implements the git server in squeak ....
Sigh, if it were really necessary to satisfy the community... One could
start from the FileSystem-Git package that ships with Squot and the Git
Browser. At the heart Git is just a key-value (or rather sha1-to-object)
store that contains blobs, trees, commits, and tags, plus a another
key-value (string-to-string/sha1) store for the refs, so the server would
even be free to choose how to persist the data.
Still, I think there are enough Git servers and implementations out there
already. We should not write our own.
If we had the time to write a Git platform, we could instead write that
integrated issue tracking platform for Monticello. The problem is already
solved for Git.
Squeak - Dev mailing list wrote
> 1. any git client interacts just as with the c git.
> 2. Behind the scenes, git uuid's are used but so are monticello "stuff"
SHA-1 hashes, not UUIDs in Git, but yeah you could put anything behind the
scenes as long as you can satisfy the interface.
Squeak - Dev mailing list wrote
> 3. Monticello interacts with the sqGit server seamlessly.( Handoff?)
So either we have that true Git adapter for Monticello (then we would also
not need our own server implementation), or the server has a Monticello
frontend in front of the Git backend.
If you were architecting this on a tight budget, which would you choose and why? If you were architecting this with an unlimited budget, what would you choose and why?
Squeak - Dev mailing list wrote
> 4. Bug happens...native squeak tools...
Unless the bug is on the server. Then hidden native Squeak tools and broken
connection...
Right. But we suffer that with Monticello too. However, Git, like Monticello, has the advantage that one can work locally without uploading, unlike, say, Subversion, which (IIRC) can only do remote commits.
Squeak - Dev mailing list wrote
> If i had time, I would do it as it sounds like a fun project. I do not
> have time. ):
Oh if only I had too much time...
Amen. But there is a corollary I can attest to. I've been able to work on Cog since 2008, 12 years already. That's as long as I worked on VisualWorks, and there's no sign that I'll have to stop working on opensmalltak-vm (Cog), Squeak or Terf any time soon. I had the architectural idea for Sista back in 2003, and had to ait for Clément's arrival to see it realised. This is to say that it is really important not to allow your ambition to be limited by your perceptio of how much time you have. Likely, if you're committed and passionate and lucky, you will have much more time than you imagine. Patience *is* a virtue, along with stubborn doggedness.
--
Sent from: http://forum.world.st/Squeak-Dev-f45488.html