Hi,
when I started getting more involved in this project I thought about
this comment from [1] a lot:
Do you *need* to have your entire history? Convince everyone to check
in everything they have, resolved all conflicts, whatever, and then
check it all in to Subversion.
Sure, for the first X days/weeks, you might need both systems, but
eventually you can just retire the entire VSS archive onto CD or DVD
and file it away.
There is a lot of truth in it. What you really need are specific paths
of development, for example your release and maintenance branches and
perhaps the state of the project at specific labels. A year after the
conversion, you wouldn't bother any longer how things started (except
for the blame) function. But would you really sue someone for an error,
that wasn't detected over more than a year?
While I was playing with the possibilities of VSS and the limitations I
got deeper and deeper into the problems and finally I wanted a 100%
conversion, simply because I wanted to prove that it is possible. The
more work you put in it, the harder is it to decide to stop. The
additional problem is, that all people who start working on such a
project, finish all their effort, after their conversion succeeded. By
nature this is a "use it once and throw it away" kind of software. If it
doesn't work out of the box, nobody will dig into the problems. Why
should they. There is a reasonable alternative by converting only the
last state of the project.
After more than one year of part time involvement in this project, I
found lot's of scenarios, where I recognized that our "release" branches
are not buildable any more, and I have forbidden my developers to
perform any rename activities in VSS. We still use VSS, since it is
convenient to use with VS6 due to the integration into the development
environment. But at least when we switch to the new vs2005 within the
next two month I want to use subversion.
So for me this is a hobby project. I tried it, I got involved, I was
soaked up by the problem domain, I spend nights and nights on
understanding the layout and finally we have a wonderful converter. If
we can solve the label handling within the next two month, I'm happy to
continue with a 100% converted archive within our development. If not, I
will continue with only the tip version converted. The existence or non
existence of a good converter shouldn't stop you from moving away from
VSS. Do it sooner than later. A 100% conversion will only make this
decision easier.
But I also mentioned it earlier. VSS isn't that bad. It does the job it
was designed for, easy version control. As long as you do use the
advanced commands, you won't have any problems. Subversion isn't also
perfect for every job. For example Linux kernel development is not done
in subversion for good reasons. If you do development on windows you
start with VSS and after 2-3 years you recognize that you need more and
that you have reached the limit of VSS. At the least at this time is
time to make a switch. Nowadays we have subversion in a usable state.
Therefor I would strongly recommend to start with subversion directly,
since it can fulfill both requirements: Easy in the beginning, powerful
in the possibilities of deployment.
Dirk
[1] <http://discuss.joelonsoftware.com/default.asp?joel.3.181938.11>
_______________________________________________
vss2svn-users mailing list
Project homepage:
http://www.pumacode.org/projects/vss2svn/
Subscribe/Unsubscribe/Admin:
http://lists.pumacode.org/mailman/listinfo/vss2svn-users-lists.pumacode.org
