Question RE subversion (web)projects setup

[Simon <gr...@gmail.com>]

Hi There,

Im pretty new to subversion but getting the hang of it. I have a setup
that we have got going which i would like to get some feedback on and
also ask a specific question please?

We have several web developers inhouse and a few outhouse. We have
subversion setup in a testing environment for a small project to test.
All users are using Zend Studio using https/webdav to
checkout/update/commit files.

Inhouse users have a personal directory which their checked out files
go on the (local/debian) web server (edited over a samba link): eg:
/home/userA/projectA, /home/userB/projectA, etc Then they can get to
this via apache thus: http://projectA.userA.domain.com/, etc for
testing their changes etc.

Outhouse users have a similar setup, but using debian on vmware on
their local workstation for personal testing.

Then we have setup /home/development/ directory, checked out the
project, then a post-commit hook does a svn update on. This seems to
work well (at this stage). The clients can get to the "projectA"
project via: http://projectA.dev.domain.com/ to test finalised
development sites.

Questions:

1). What is wrong with this? :)

2). We have a nice little sub 1MB project as projectA, im concerned
when we then go to load in a couple of GB's of projectB, C, D etc..
what will happen here?

3). Further to above... should we be having separate "subversion's"
for projects rather than lump them all together?

Sorry if my terminology is not quite right, but helpful hints and
suggestions are very welcome. We all LIKE what we see so far!!

Thanks

Simon

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Question RE subversion (web)projects setup

[John Peacock <jp...@rowman.com>]

Simon wrote:
> Questions:
> 
> 1). What is wrong with this? :)

Nothing - it is how we work as well.  We have two servers (one production and
one test) both updated by post-commit hooks, so that the developers don't need
to do anything special (this is a good thing).  I even wrote a Perl module to do
this easier:

	http://search.cpan.org/search?query=SVN-Notify-Mirror

so internally here, the test server is synced on every commit and the production
server is synced when a specifically named tag is created.

For your external developers (I think "outhouse" is not the right word ;-), you
can either have your test sites available on a public IP address (as you do now)
or use SSH tunneling to bring the external users into a private site.  You can
even have a third, staging server, which is externally accessible by the clients
for final approval

> 2). We have a nice little sub 1MB project as projectA, im concerned
> when we then go to load in a couple of GB's of projectB, C, D etc..
> what will happen here?

It works just fine (on the Subversion side), modulo the requirement to have
multiple working copies checked out on multiple machines.

> 3). Further to above... should we be having separate "subversion's"
> for projects rather than lump them all together?

A good rule of thumb is that all related, or likely to be "borrowed", code
should be in one repository.  If your client files are 100% independent, it is
better to use separate repositories; for example, if a client leaves, you can
backup their repository to DVD and delete it from your server completely.

HTH

John

-- 
John Peacock
Director of Information Research and Technology
Rowman & Littlefield Publishing Group
4501 Forbes Blvd
Suite H
Lanham, MD 20706
301-459-3366 x.5010
fax 301-429-5747

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org