On Oct 5, 2005, at 09:55, Turaukar Yur wrote:
> I was looking at various version management systems to store
> information. As a software development team, we have the need to
> store and manage documents (specifications) and other binary data
> as well as source code.
>
> Being well aware of subversion and its focus, I ran accross Apache
> Jackrabbit - an open source implementation of the Content
> Repository for Java Technology API (JSR-170). While that project
> has not yet released any downloads yet, it seems to be very
> interesting as a storage engine for document repositories.
>
> The question for me here is: do I need two systems or could
> Subversion do the job alone? What's the relation between the two?
I don't know any more about Jackrabbit and JSR-170 than what you've
just told me, but I can answer these other questions you had:
> From a first glance, Jackrabbit is focused towards handling large
> documents and binary data, as well as some XML support -- the
> latter which I frankly didn't understand.
> Subversion is focused on handling source code -- presumably smaller
> text files -- but can basically handle binary files as well.
> Anyone knows about restrictions/limitations in subversion? Is it a
> good idea to manage a large number of binary files through it?
Subversion can handle binary files of any size just fine.
There is a 2-GB limitation per file when using APR 0.9. This also
applies to Subversion's revision files, meaning that if you try to
commit three 1-GB files at once, you'll create a 3-GB-large revision
file, which won't work with APR 0.9. Use APR 1.0 if this is important
to you.
> Jackrabbit also supports searching through the documents -- by
> content and attributes. What are Subversion's features here? How
> efficient is searching by attributes? Is there any "plug-in" that
> provides fast index-based searching on document contents?
AFAIK Subversion's attributes are not searchable, with a few
exceptions—you can find a revision by date, for example. But you
can't search arbitrary properties, and you can't search content. If
you want to search content, you'll have to make a working copy of the
part of the repository you want to search, then use other tools (such
as your operating system's file searching capabilities) to do the
search.
When you make a working copy of a Subversion repository, be aware
that the hidden .svn directories will contain a complete copy of the
data. If you're versioning large binary files, the additional disk
space this uses may be relevant. The ability to have a working copy
without this pristine copy of the data does not yet exist, but
feature requests have already been written, and I believe someone may
already be working on the implementation.
You can of course to an export (instead of a checkout), and you
wouldn't have the .svn directories and the associated space usage.
But you also wouldn't have a working copy then, meaning if you make
any changes, you can't check them back in.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Oct 5 11:13:46 2005