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

progress report & upcoming "Milestone 1"

From: Karl Fogel <kfogel_at_galois.collab.net>
Date: 2000-09-26 03:53:06 CEST

Just a brief update on where we are and where we will (hopefully) be

Milestone 1 is due on Oct. 7th. This milestone has been sometimes
known as "Hello, world", but somehow that name feels like it implies a
greater amount of active networking code than the release will
probably have, so let's just call it Milestone 1 instead.

What does Milestone 1 mean?

   - Checkouts of complete trees
   - Updates of file changes, and file adds
   - Commits of file changes, and file adds

These features will work in both the client side and the server side
libraries. However, it probably will not be feasible to have the two
sides actually talking to each other over the network in this release,
because that would require hooking things into Apache, which will be
mildly hairy. Since it makes more sense to have these features solid
and debugged in the libraries, we'll concentrate on that, and then
immediately after this milestone we'll work on hooking everything into
Apache, so that future testing is really happening over the network.

For this milestone, we'll just test by hand-creating xml tree deltas
in files, and feeding them to the client and the server. That's how
we've been testing up till now anyway.

I should emphasize that deferring the network-hookup is not in any way
representative of progress in writing the networking layer itself,
Greg Stein has been doing that right on schedule (even ahead of
schedule); the thing being deferred is actually connecting his modules
with Apache and with the client.

As for progress:

Jim Blandy is hammering away at the filesystem code, despite some
unavoidable interruptions (he still retains some non-Subversion
responsibilities at RedHat while he's working on this project). We
talked on the phone today, he seems comfortable with the Oct. 7th
milestone, given that it's about the library interfaces themselves,
not about hooking up the networking code.

The working copy library passed a "milepebble" today, in that
checkouts are now producing valid working copies with the exception of
properties. (The main thing that makes this a milepebble is that the
versions file is now correct at the end of the checkout.)

All of this could have been accomplished a month ago, by faking it;
the nice thing about this pebble is that the checkout is happening in
the intended crash-proof way. So none of that code should need
rewriting now, except for bug-fixes, but of course there won't be any
of those. +thud+

And as you have seen from checkin messages, Branko and Greg Hudson
have basically written the vdelta library. Wow. Thanks, fellas.

What to do now:

I've added a bunch more tasks into the TASKS file. Please look them
over. Whatever time you can volunteer is most appreciated. And if I
may be so bold as to suggest: Branko and Greg H please take a look at
item #3, which I think you guys might be most qualified for.

Okay, that's it for now. Back to coding...

Received on Sat Oct 21 14:36:09 2006

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