Anatoly, the main usability problem with GoogleCode (as Joseph
mentioned) is the lack of a good pull requests interface. That's what's
making our new potential contributors (like Joseph himself and Thomas
Kluyver) to stay away from doing it.
I think there is no other way to address this issue than by moving our
main repo to another service, because GoogleCode is showing not signs of
improvement.
El 14/11/13 03:29, Joseph Martinot-Lagarde escribió:
Le jeudi 14 novembre 2013 01:15:27 UTC+1, anatoly techtonik a écrit :
On Wednesday, November 13, 2013 4:53:56 PM UTC+3, Joseph
Martinot-Lagarde wrote:
The discussion on
https://groups.google.com/forum/#!topic/spyderlib/5tw2ZItlxUM
<https://groups.google.com/forum/#%21topic/spyderlib/5tw2ZItlxUM>
remind me of
http://code.google.com/p/spyderlib/issues/detail?id=816
<http://code.google.com/p/spyderlib/issues/detail?id=816>.
Google code clearly lacks functionalities compared to
Bitbucket and Github (the main one being pull requests, I
think). In addition to this is the eventuality to shift to git
instead of mercurial.
Both GitHub and BitBucket has awful interface, which is confusing
for new users, who may not be Github clients. I am speaking about
wiki pages. Issue trackers are also very limited compared to
Google Code.
Well it's a matter of taste, I don't like google code interface at
all. Wiki pages is a minor inconvenient (if any) compared to the
usability of pull requests.
I had problems to write the first post here: on each new line if would
switch back to bold. Yay for the interface ;)
From the tip of my head, here are the pros and cons I can find
for each service :
*Bitbucket/Mercurial
*+ Uses mercurial and git. This allows to keep mercurial as VCS.
+ TortoiseHg
- less users
+ simple
+ free private repositories
True, but private repos is not a concern for spyder.
*Github/Git
*- Git only
+ numpy, scipy, ipython and matplotlib use it
+ more users
- tracker data has a proprietary format (but is it important ?)*
*
There is also the possibility to have read/write mirror I
guess, but I have no clue of how it works...*
*
Pull from Bitbucket, commit to Google Code, and it is synced
around. No read/write mirror is possible without auto merges.
Why I prefer Git over Mercurial :
+ 2-stage commits helps to check the correctness of commits
2-stage commits really suxx. Use `hg record` if you're unsure
about what you're committing.
It sux for you maybe, but it is very logical to me to first choose the
modifications you want to commit, then validate the commit. It fits my
workflow very well. Thanks for `hg record`, it seems to be a direct
equivalent.
+ easy selection line by line or block by block instead of
whole files for commits (using git gui)
'hg record', no GUI required.
GUI is not required, it just makes it easier. I usually prefer a good
gui to any command line. (Sadly good gui is not that common.)
+ git stash
Mercurial queues or "hg diff > stash"
--
You received this message because you are subscribed to the Google
Groups "spyder" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/spyderlib.
For more options, visit https://groups.google.com/groups/opt_out.
--
You received this message because you are subscribed to the Google Groups
"spyder" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/spyderlib.
For more options, visit https://groups.google.com/groups/opt_out.
