[Biopython-dev] history on github - where are the tags?

[Peter]

On Wed, Apr 22, 2009 at 9:53 AM, Bartek Wilczynski
<bartek at rezolwenta.eu.org> wrote:
> Hi,
>
> Peter wrote:
>>> There is another option, redo the import using git cvsimport.  This>
>>> has the downside that we lose all the network history currently in
>>> github, but its only going to affect a couple of people and that was
>>> always a possibility.
>
> Yes, it is  an option, but I would be quite reluctant to do it. I think this
> issue with tags is possible to get fixed without re-doing the import.

If you can fix the current git hub repository, great.

> I'm scared by the possibility the we re-import stuff, fix the tags, everybody
> swithches, people complain how good it was back then with CVS, ane one
> month down the road, we find that there is an issue with something else,
> that was not present in the previous import.

This is why we are testing things: We have found something wrong with
the current import, and it wasn't immediately obvious (partly because
we were still getting to know git and github).

> I think this is becoming a bit chaotic now. We still haven't removed the
> first github conversion: (biopython_old branch: is anyone using it anyway?),

The old conversion's deletion is still in progress, it must have stalled:
http://support.github.com/discussions/repos/485-reposiotry-stuck-in-rename

> there is this semi-official one that has a (fixable in my opinion) issue with
> tags ...

If we can fix the tags, great.  If we can also remap the authors to
their git usernames, even better.

> ... and now there is a new one made by Peter.

I deleted that one - it was just a proof of principle.

> In summary:
> I have no objections to using any particular tool for importing stuff to git.
> I don't like the idea of not even trying to fix the problem we have
> but instantly changing the tool we are using.

It was really to demonstrate to my own satisfaction that we could have
the tags in the history properly.

> I consider now re-importing stuff a major problem: everybody will need to port
> their changes which is work.

True - but this was always a possibility.  From browsing the github
network this really will just affect basically just two people:

* Eric - quite a few changes, some of which we can probably look at
merging into CVS now which would solve that.
* Giovanni - quite a few changes (on a couple of files) on one branch,
and a couple of other branches for proposed unit tests

Also:
* Dave Bridges - documentation changes to one file which we can merge
into CVS and then he can delete that branch
* Tiago - trivial changes to one file (stats in PopGen)
* Peter (me) - I have a few test branches, nothing I care about.

Brad, Bartek and Leighton have no changes made.

Peter