Fabien COELHO wrote:
>>> (2) I use this ASF repository as an example of what should not be done
>>> with svn when I teach svn:-) I advocate that pretty independent
>>> should be in independant repositories, and connections can be
>>> with svn:externals quite simply.
>> Heh. Well, I teach people to avoid svn:externals like the plague. :)
> Differing opinions, good:-)
> I did not have much trouble with externals up to now, and I use them a
> lot. The only issue is that if we want to do a global tag or branch on
> related projects, the references must be updated accordingly, so we
> scripted it.
> So, everybody wants a very big repository with hundred of unrelated
> projects put together and dead references everywhere in the source and
> outside... and the strangeness of this very discontinuous choice does
> not strike anyone but me. Well...
Ha. So, from a strictly CM-academic point of view, certainly, we should
keep our history intact during transfer -- and that includes keeping
revision numbers and implies not grandfathering in the CVS repository.
I'm leaning the other way because a) we'll get a *lot* more serious
testing of release candidates; b) we'll actually deal day-to-day with a
repository that's orders of magnitude bigger than our own (death to
scalability problems!) and c) it will give /us/ a bit of incentive to be
extra careful checking the quality of release candidates.
Sometimes you just have to be pragmatic and break the world for everyone
so that you can build a better one. ;)
Received on 2009-11-11 20:24:05 CET