-
-
Notifications
You must be signed in to change notification settings - Fork 1.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Transition to git and github #1911
Comments
From contrebasse on 2014-07-31T04:04:07Z It was discussed when swithcing from googlecode to bibucket. The main reason for not swithching to git is that the core developers prefer mercurial. https://groups.google.com/forum/#!searchin/spyderlib/github/spyderlib/F0N3MdUNEtA/cP1DmEmibssJ There is a previous discussion I just discovered: https://groups.google.com/forum/#!searchin/spyderlib/github/spyderlib/z4JrnmAE3rA/oPUp8_Y8hI0J I personnaly prefer git and github, but I'm quite happy to contribute using bitbucket! The separation of the code and the issues is a problem though. |
From jfilippi...@gmail.com on 2014-08-01T10:49:54Z Thank you for linking to the relevant discussions. I will try to find the time to read the arguments there and potentially we can revisit the matter. |
From ccordoba12 on 2014-08-01T11:15:45Z I really like hg better, but here are my answers to your points:
Summary: Transition to git and github (was: transition to git and github) |
From ccordoba12 on 2014-08-23T14:03:52Z I'd say that unless you put the time and effort to do the migration yourself, this is not going to happen anytime soon, sorry. |
From jfilippi...@gmail.com on 2014-09-06T01:24:35Z Sorry for the delay in answering, but I wanted to first read the threads posted by @-contrebasse.
I don't know about threaded talks, I think github's linear issues are much better to follow and revisit a long time later. Threaded conversations are for mailing lists, not for public issues. There hasn't been a contributor that has complained yet because they just dismissed they idea w/o discussing it, or they didn't even consider it in the first place. The reason is that many contributors are people who will send misc pull requests because they happened to find (and fix) a bug, or implement some small new feature. They do not plan to make a big commitment: if it already fits their workflow (git, github etc), then they will send it, otherwise they may as well be discouraged. Also, there are the major contributors to the core scientific python projects (numpy, scipy, matplotlib, ipython, networkx,...). The fact that they are on github and using git minimally tells us that they approve of that approach better, and may prefer to not contribute in another way. Attracting them will be a benefit. Dismissing the implications of their decision to use github is not easy, and cannot be based on comparing only hg to git, or bitbucket to github.
|
From jfilippi...@gmail.com on 2014-09-06T01:34:34Z From what I find there already is a synced mirror on github here: https://github.com/spyder-ide/spyderlib-mirror and an older experiment here: https://github.com/ccordoba12/spyder By migration do you mean transfer all the issues, or just convert the repository (which is trivial with hg-fast-export) ? I would be happy to convert the repo itself (although I guess that is not what you refer to). About the issues, I could try: https://github.com/arthur-debert/google-code-issues-migrator (or maybe instead: http://trentm.com/2012/03/google-code-to-github.html ) I don't think though that the issues need to be transferred. It may be better to move the code there, and direct there for opening new issues, while issues here are being closed. This will make the move smoother. I understand that cross-referencing may be somewhat inconvenient for old issues, but you can still simply link to them. Also, if the script linked above is actually used and proves functional (from its popularity I expect it will), then cross linking should work fine. And I'm sure there will exist some straightforward solution to edit git history so that any commit messages referring to hg commit hashes are replaced with the git commit hashes (this concern was mentioned somewhere in the discussions linked above). |
From ccordoba12 on 2014-09-06T08:42:09Z I was the first one to propose we should move to Github about 3 years ago. Since at that time my proposal was dismissed, I got more and more accustomed to hg and now I'd find really hard to do the switch :) However, I find some of your reasons valid:
So please bring this issue to the mailing list to see what other people think about it. The only thing that worries me a bit is doing another transition in less than a year, which could be quite confusing for users and contributors alike. |
From jfilippi...@gmail.com on 2014-10-01T23:04:33Z Thank you. The suggestion seems to be supported also by others, based on the suggestion in: https://groups.google.com/forum/#!topic/spyderlib/VH6SRGYwt-E The transition should not be very difficult, in particular if the old issues remain on googlecode and any new issues are redirected to github. In any case I run a small experiment, transferring the first 35 issues from googlecode to github: https://github.com/johnyf/spyder_experiment/issues?q=is%3Aissue+is%3Aclosed using this python script that was cited also earlier above: https://github.com/arthur-debert/google-code-issues-migrator Some helpful other remarks can be found on this blog: http://trentm.com/2012/03/google-code-to-github.html |
From ccordoba12 on 2015-02-08T11:33:52Z Labels: MS-v2.3.3 |
This has been done, as you can see ;-) |
Oh yeah. Thank you so much for this. New contributions coming soon. |
@SylvainCorlay really great to know!! |
From jfilippi...@gmail.com on 2014-07-28T17:43:22Z
This is a suggestion that most probably has already been considered, but I would like to strongly recommend it. The reasons are:
Original issue: http://code.google.com/p/spyderlib/issues/detail?id=1911
The text was updated successfully, but these errors were encountered: