Evaluating Bazaar in a CVS environment.

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Andy Walker wrote:
| Alaa Salman <alaasalman <at> terra.net.lb> writes:
|
|> But so, say you want to test your changes... Having only a few files
|> that you want to modify would not allow you to do so. How would you
|> build the project?
|>
| That's a very good question.  But what of the case of php web
applications or
| applications built on perl that don't require "building" per se?  They
only
| require that the latest version of a particular file be in place.
|
|> Plus, if you don't want to waste space for example... host the "main"
|> branch on some server, and then do a lightweight checkout and hack on
|> it. I guess this would be working in a similar fashion to how a
|> centralized revision control system would work.
|>
| Yes, this would save space on the local machine initially, but the
main issue I
| was talking about above would be avoiding duplicating a source tree
with many
| thousands of files if you only need to change 6 files in three separate
| directories.
|
| This would be ameliorated somewhat by splitting these directories into
their
| own logical units as mentioned above, but then it gets into the other
issue of
| needing to modify files in three directories-projects of equal level
for one
| particular task and the ability to group these three separate(?)
branches into
| a logical unit signifying that they come from the same task.
|

The idea that I presented was to allow you to have 1 lightweight
checkout, that you can switch between branches. So you only have a small
number of copies (I personally like to always have a checkout of HEAD,
and then one for what I'm working on.)

But I have 30+ branches on my local disk, and in total they only take up
~100KB of disk space. Versus the 20MB working tree checkout.

|> I believe that you're still thinking in the CVS terms of files, while in
|> a DVCS like bazaar, you need to think of branches as sort of a single
unit.
|>
| No, I have a decent idea of HOW it's supposed to work and of the
notions of
| branches etc, I was just under the impression that many different branches
| could operate off of the same working tree without requiring a separate
| checkout (i.e. an entire separate copy of that entire section of the
source
| code) for each branch.

I did present that, I guess not well enough for you to understand.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkhlXnwACgkQJdeBCYSNAAMCuQCgsHsefqaUt46tdUD1MRej3lZV
DMAAnRb3Bd1zYxOflgCYmjvMc6txGjgc
=t36U
-----END PGP SIGNATURE-----