All,
I manage a farily large CVS repository as well as
an RCS repository (don't ask! :-)
The CVS repository is around 100,000 files and
supports a large organisation.
Having studied many versioning systems free and some commerical
I currently think Subversion is a pretty good version model.
I have read through the Subversion manual and
watched the mail groups.
I have some concerns that someone might light to
comment on:
* Several free repository version systems (eg PVCS)
seem to have died a silent death. Subversion
whilst it seems fairly active and apparently quite
well advanced could end up the same way.
Also Subversion is very much Alpha (even though
the database is well established).
"Convince me" that Subversion is here to stay
and that it is being proactively developed
by a broad enough base of developers to make
it self-sustaining.
* the CVS conversion tool aparently has a few niggles.
"it doesn't do branches or tags very well yet"
"there may be tiny bugs in the way it deduces CVS
commits."
Perhaps someone could detail these more accurately
Also how well will it preserve my tags, vendor
branches etc.
* Does Subversion's checkin re-do the deltas?
I effectively checks out version 1, version 2 and
does it's own diffs then does version 2, version 3
etc ?
* When importing from CVS will Subversion turn
non-delta binary checkins into binary deltas?
(ie CVS doesn't do binary deltas)
* How should I import an RCS repository
* I understand why a database has been used instead
of a directory tree (ala CVS). Nevertheless
it gives me the "screaming hebe-geebies"
to think of committing my 100,000 files to
a black-box repository never to be seen again.
You can see where I'm coming from I'm sure.
I also have to convince others with similar views.
Again "convince me" :-)
* Is there yet an option to not have original copies
in the .svn directory. We have several "packages"
that contain large binary files. I don't want
to have extra copies in the .svn directory.
I understand the limitations of this but in
some cases these "hidden" extra gigabytes of
disk can also be an issue...
* IMHO good CVS conversion tools are probably more
important that most things other than stability and very
basic features - ie I don't think you should wait until
version 1.0 to finish these.
My reason is that converting from CVS is probably
your biggest user base. People won't convert and
build up a big Subversion userbase unless conversion
is 99% reliable.
Comments ?
* And the oft asked question I'm sure
"Is there an approximate timeframe for version 1.0.1"
(never buy version 1.0 of any software :-)
Yes I know this is hard but #features*time per feature
divided by #fulltime (?) developers -> ???
Could one expect 6-12 months, 12-24 months, 3 years ?
I'm just trying to get an idea of people's/Tigris'
commitment to the project -eg full time developers
for a year or two equals commitment. 50 People doing it
in their free time is also reasonable commitment.
5 people in their free time is cause for concern.
Don't get me wrong - I'm not implying anything negative
and I think Subversion is a great leap forward, but I
also have to consider management concepts such as
"corporate profile", "investment", "product development
path" etc. Putting source code worth multi-million
dollars into someone's "pet repository project"
would not go down too well.
BTW I wouldn't be asking if I wasn't already
convinced that Subversion is worth looking into
seriously.
regards
Kim
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Mar 26 09:37:03 2003