On Fri, Jul 2, 2010 at 10:17 AM, anatoly techtonik <techtonik@gmail.com> wrote:
On Fri, Jul 2, 2010 at 4:53 PM, Georg Brandl <g.brandl@gmx.net> wrote:
What is the problem with realtime synchronization and working with already up to date Mercurial mirror of central SVN repository?
The specifics of the conversion process are not nailed down yet. Therefore, the exact translation of SVN to Hg commits will change, and with it the Mercurial revision IDs, for example.
Does anybody here know how Mercurial calculates the IDs? From that I remember it is author + message + diff content. What can change there?
Development will continue in SVN repository until everybody is ready for final migration in X weeks later. Is that right?
No; as soon as we switch, SVN will be read-only.
Why don't allow people who already know Mercurial use Mercurial and those who prefer Subversion use that. If Mercurial allows to submit to Subversion - why people can't use that while we writing tutorials and answering question about workflow?
I don't think that we have enough manpower to maintain such a bridge indefinitely.
It doesn't require manpower. It requires automation. Considering that the biggest problem now is to get sane lossless conversion, we should elaborate on getting this in place. After that I would still follow the path of setting realtime mirror for X weeks that could be replicated by bitbucket, launchpad and other services to see how people pickup the changes.
As PEP 384 says - the transition is mostly to make lives of outside contributors easier. Core developers can wait for a while.
No, we can't. We *have* been waiting.