Firstly a not so short intro...
I come from a slightly different community to that of the Sorceforge or
other open source communities, thought some overlap exists. My
interests are academic and I develop code for research in astrophysics
and mathematics.
CVS is (slowly) being picked up by groups as they realise that it is
very useful in local/global collaborative projects or even just version
controlling your own work, which may be picked up by somebody else
later. The academic research world is perculiar in that it has been
managing global collaboration for a very long time (i.e. longer than the
commercial world). Collaboration it is considered by many to be that
which describes the academic world.
Uses of CVS
-----------
1) Developing software codes (D'oh)
2) Histories of runs of numerical models
(eg. run parameters and outputed time series)
3) Writing papers (it is quite efficient when writing in text based
languages such as LaTeX, with postscript figures)
4) Sharing libraries (more static than active codes) of functions,
visualisation routines, configurations for numerous languages and
applications
5) Maintaining web sites
Why CVS?... It's free and it's the academic world.
Some of the security features are used to allow particular people access
to current projects but generally is is only the very basic functions of
cvs we tend to use. Many users are primarily mathematicians and
computer ops second so co, commit, update, add, remove, diff are almost
all that is ever used. Branching and tagging appear too 'complex' for
most!
-------------------------
Anyway, I am in the position of trying to encourage various people in
our department to use version management as many of our international
collaborators do. I have taken some time to consider SVN and come up
with a couple of difficulties of which I would like to make you aware:
1) Many of our users (students esp.) do not have vast amounts of disk
space, as your disk space vs. network bandwidth. Quotas are
generally tight in academic institutes unless good reasons
are given. Trying to persuade them to keep two copies of everthing
they check out is almost impossible, especially when they really
only wanted to check out a read-only copy to peruse!
Further we've all got network bandwidth coming out of our ears :-)
2) CVS is capable of having 'local' repositories. i.e. even if you
don't have server you can wack in a directory, cvs init it and
use it for version control / internal collaboration. SVN can't
do this... can it? Local repositories really are useful, especially
in situations where a code may have many developer chronologically
though one a poor postdoc and his supervisor at any one time. A
local repository remains an easily passable parcel, but one can
still grep the cvs log to find out when/why something changed!
(I know, I don't love this either, but it really is how people work)
3) One of the great advantages of CVS is when you develop and test
locally then run codes on a super computer... Since you're working
in CVS one can simply point the CVS at your repository (e.g. here
in Newcastle) check out the code on the super computer and
compile it. Unfortunately I have yet to found a super computer
with Subversion installed, hopefully this will come with time and a
little. Is it in any major distributions yet?
tOnY
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Mar 11 18:50:06 2003