[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Questions from the Subversion 1.6 Preview event in Berlin

From: Hyrum K. Wright <hyrum_wright_at_mail.utexas.edu>
Date: Wed, 25 Feb 2009 16:18:02 -0600

On Feb 25, 2009, at 3:31 PM, Neels Janosch Hofmeyr wrote:

> Hi all,
> Tuesday saw the first day of the Subversion 1.6 Preview (in Berlin),
> organized by CollabNet and elego. Everything went well, and there
> were a lot
> of questions about new features and the immediate future plans.
> The committers present were Julian Foad, Stephen Butler, Stefan
> Sperling and
> me, and there were 20-odd attendees with corporate background.
> So I'd like to pass on some feedback and am hoping for comments that
> we can
> forward back to those guys who asked the questions.
> wc-ng
> ~~~~~
> There was a sort of eerie concern about wc-ng, in that it will
> change some
> basic properties of working copies on disk. There was mention of
> some people
> zipping whole working copies and sending them by Email, or storing
> them away
> for later committing. Our general reply to those remarks was "use
> branches",
> but people were insistent that some users will not be happy with
> making the
> use of working copies any more complex by adding restrictions.
> Needing to
> announce the rename or deletion of a working copy to a central
> database
> could, in essence, break a lot of people's workflow, blocking an
> upgrade.
> After stsp broke down the main reasons for and benefits of wc-ng
> (developer
> sanity, speed, concurrency, complex queries), there was some sort of
> "ok, we
> understand now". But, ...

Users with WC-NG won't need to announce anything to a central
database. As Bert already mentioned, the database will be central to
a particular working copy as a default, and the working copy will be
able to be shipped around.

> user participation
> ~~~~~~~~~~~~~~~~~~
> ... directly from that followed an almost accusingly toned plea to
> let the
> users have a say in the planning of such features like wc-ng.
> Obviously, it
> sounds outright ridiculous to say that to an open source community.
> But the
> reality is that the typical subversion user is not able to keep up
> with the
> dev@ mailing list to filter out any significant changes emerging.
> Maybe most
> of the plans aren't even written down anywhere, but are incubating
> in the
> developers' heads.

We've got quite a corpus of notes on wc-ng, including a development
plan and a meta-issue in the issue tracker. You can point people to http://svn.collab.net/repos/svn/trunk/notes/wc-ng-design
   Most other proposed features are likewise doc'd ... somewhere.

Along the lines of not paying attention to users, did this arise from
the "Speeding up workspace" thread on the users list, perchance? This
is open source, people: if you want to have influence in the
community, contribute. Period. (Caveat: "contribute" does not mean
"write code". It could mean write code, but it could also mean join
in the development discussion, fund a developer to work on your
feature, or contribute financially to SVNCorp. But being visible in
the community will give you a much louder voice when it comes to
features, etc.)

> So maybe we need some justification for and an estimate of
> consequences of
> (particularly wc-ng, and maybe other) changes that are still in the
> pipeline, published on tigris.org, easy to find.
> And an affinity to officially released prototypes for user feedback
> was
> mentioned.

I used to have a nightly script running on orac which would create
proto-tarballs; I'd be happy to resurrect it for trunk. The only
caveat is that people need to know that it's completely untested code,
with all the caveats that it implies. As a developer, I'm happy to
have people running this code, so long as they understand it might eat
their children.

> other questions
> ~~~~~~~~~~~~~~~
> - We were again asked about svn obliterate. How far are the plans?
> Is it going to come any time soon?

There's been some rumbling on the list about designs and such, but
zero implementation effort thus far.

> - We were again asked about server-imposed rights restrictions.
> This time the wish was to "at least" be able to forbid certain
> operations in certain directories. Our gut answer was:
> use commit hooks. Anything to add here?
> - We were asked about automatic commits of svn:externals (directories)
> together with their parent directories in the working copy. Can this
> be switched on somehow?

Are people concerned that this is happening? Or is this a feature

> - And interest was mentioned in *sparse* svn:externals, to be able
> to include a directory to a limited depth as an external.
> So, here's to us to answer the outcry of our corporate users. I'll
> gather
> our comments and forward them directly to the participants of the
> Subversion
> 1.6 Preview, labeled "comments from the dev@ list".

Thanks for the writeup. I generally appreciate this type of feedback
from the user community, just so long as they aren't complaining *too*
loudly. :)


Received on 2009-02-25 23:18:27 CET

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.