RE: Partial checkouts on a huge project

You can do it the hacky way by

1) Do a non recursive checkout of /trunk.
2) Edit trunk/.svn/entries
2a) Add a directory entry for tools: <entry name="Tools" kind="dir"/>

3) Do a non recursive checkout of tools into trunk/tools
4) Edit /trunk/tools/.svn/entries
4a) Add a directory entry for common: <entry name="Common" kind="dir"/>

5)Do a non recursive checkout of common into trunk/tools/common.
6)Edit /trunk/tools/common/.svn/entries
6a)Add a directory entry for tools: <entry name="drill" kind="dir"/>

Go back to the trunk and "svn up" will update only the directories and files
you've checked out.
It feels hackly, but it does exactly what you are looking for.

-----Original Message-----
From: mike nicholaides [mailto:mike.nicholaides@gmail.com]
Sent: Friday, April 28, 2006 11:05 AM
To: Gale, David; users@subversion.tigris.org
Subject: Re: Partial checkouts on a huge project

Yeah, I thought about recursively doing non-recursive checkouts, but
that's a pain.

Another problem that I forgot to mention was that when you I have a
huge project, svn status and commits take a really long time. I was
considering making a super-simple web-based mechanism to view the
status, commit, update, etc. So, I would ideally want it to be a
quick operation.

Well, if I can't find an elegant solution for the my problem, I will
do what you suggested, David. Thanks.

Anyone got any bright ideas?

Thank you all,
Mike

On 4/28/06, Gale, David <David.Gale@hypertherm.com> wrote:
> mike nicholaides wrote:
> > Hey guys,
> >
> > I hope y'all can help me out. I've been googling and reading the
> > archives of this list for about a week now.
> >
> > I have a huge intranet project that I want to put under version
> > control. The problem I've run into is that the whole project is 1.3
> > GB, which means that checking out the whole project, even on the same
> > machine as the repository, takes forever.
> >
> > Checking out sub directories won't work, because they rely on the
> > parent directories. Non-recursive checkouts of directories doesn't
> > work either, because I can't recursively commit from a non-recursive
> > checkout.
> >
> > Here's a simple example of what I want to do, in case the previous
> > paragraphs are vague.
> >
> > I have a directory structure:
> >
> > trunk/
> > tools/
> > common/
> > hammer/
> > drill/
> > chisel/
> > custom/
> > cde24/
> > abc92/
> > customers/
> > projects/
> > ...
> >
> > Now, say I want to work in "drill" directory. The project depends on
> > files in the parent, grandparent and ancestor directories.
> >
> > So, I want my check out to look like:
> > trunk/
> > tools/
> > common/
> > hammer/
> > And, I want trunk, tools, common, and hammer to have all their files
> > checked out, but none of the directories (except hammer should have
> > all child directories checked out).
> >
> > Does this make sense at all?
> >
> > What do I do?
>
> Standard advice is to do the initial checkout (which, as you say, will
> be fairly hefty); after that point, you shouldn't need to do any full
> checkouts again--just updates, switches, etc., which will all be fairly
> fast.
>
> Alternatively, you could do a non-recursive checkout and then update
> each sub-folder individually, spreading the network load across several
> commands, but this obviously requires more baby-sitting.
>
> -David
>

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