> Subversion doesn't preserve permissions or ownership information; it
> simply provides a generic metadata mechanism for users to exploit
> however they wish.
>
> Ben Reser, I believe, has written a script to do what you're looking
> for: when you run it, it "stores" filesystem metadata in the
> properties before committing, and also has the ability to read the metadata and
> apply it to the files again.
Many people have written similar scripts,
for non-SubVersion systems:
e.g.
* finding all symlinks,
creating a script that creates exactly
that configuration of symlinks
* similarly, recording all file
permissions and ownerships
(for files where it matters)
creating a script that creates exactly
that configuration of symlinks
I've been thinking a bit about such scripts,
and whether I want to port them to SVN.
Similarly, I've been thinking about .cvsignore
vs. SubVersion's svn:ignore property. My Makefiles
now have a target that creates propsets from
.cvsignore - not vice versa.
Why go from the .cvsignore file to the svn:ignore
property? Why not vice versa? Isn't SVN metadata
better than placing the same information in files?
Well, you know, I don't think so. At least not yet.
In the most important aspect:
SVN metadata is versioned.
Files are versioned.
But there's a secondary aspect:
if you are using SVN metadata,
the information cannot easily be shared
with other people who may not be
using SubVersion.
Whereas if this information is stored
in files, it can easily be shared with
other non-SubVersion users,
and non-SVN tools can be used to
maintain it. E.g. I could use EMACS
to do a global search/replace for
file ownerships.
Sharing with non-SVN users is something
that matters to some of us, who have
to track local changes to software from other folk.
(Anyone want to track Linux from BitKeeper
into Subversion?)
===
"Tracking metadata" was one of the big things
SubVersion was reputed to provide to CVS users.
However, I am beginning to think that it is not
the versioning of the metadata that is important:
rather, it is the automatic extraction of metadata
from files, metadata such as ownership, permissions,
etc., with mechanisms for indicating when such
tracking should be done, and mechanisms for inheriting
such control.
Given the extraction/application scripts, it is not
clear to me if there are any advantages to storing
the data as SubVersion metadata.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Sat Apr 10 03:42:02 2004