Hello Jamie!
I'm cc'ing this to the dev@-list, so other interested people see this (and
maybe help, too).
On Wednesday 30 March 2005 21:53, Jamie Kirkpatrick wrote:
> As I have such a keen interest in seeing this happen, I was wondering
> if I could help you out at all? I am happy to help in the
> development of the branch and to try to fix bugs etc. I'd also like
> to know what your feeling on whether this will get merged or not is?
> I think that the two arguments I have seen against this dont hold
> much water. In terms of it being outside of the remit, I dont see
> why it cant be an option that you can enable on a particular
> repository (or subdir) so that those who need it can take advantage.
> In terms of planning, I guess it just takes a few of us to try and
> flesh out ideas and to participate with the main development team as
> closely as possible.
Thank you for your interest.
I too believe that it should be done in svn directly and not some wrapper
scripts; and my dream is that svn gets some other libsvn_wc, which works
without .svn directories. But that will take a while.
Yes, you can help - of course.
Some of the opposition comes from the fact that this feature was done without
designing it first (which is explained by the fact that the developers had
other things to do [eg. release 1.0], and that I didn't have a complete
overview in svn).
What needs to be done?
- A description, *how* the branch currently operates, is missing,
I'll write something in the next time, and push it into the
meta-data-versioning directory.
- Some design considerations have to be made regarding conflicts:
what happens if a file has an unix-mode of 0700 [checked-out],
locally changed [via chmod or "svn ps" ?], and would get updated
to another mode? Which should be taken? The other properties, too.
Here some analysis and debugging will be needed, too - what is the
current behaviour?
- What I'll probably need in the next months (and you perhaps too :-)
is that other special files, like character and block devices, get
versioned. Fifos and sockets are not a problem, because they'll likely
get generated upon program (daemon) start.
- Perhaps somebody would like to use that on Win32 too.
There we'll have the problem that there is *no* good way to specify
the ACL's as readable texts (allow/deny ACL's, a lot of bits with varying
meaning, inheritance, local/domain users, ...)
> anyway, I'd like to hear what you think about what I have said....
>
> Hopefully there is something i can do to help this all move along..
So there are some things where you can help. Send patches to dev, and cc me;
I'm certainly interested.
> Cheers
>
> Jamie
Thank you for your offer!
Regards,
Phil
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Mar 31 07:16:34 2005