On Sun, 21 Sep 2003, Daniel Vainsencher wrote:

> > Is there anything else that people would like added to ViewMC?  Is it at
> > all useful?
> Well I haven't used it yet, really. What do you use it for/did you
> intend it for?

Largely as a way of browsing the history of a package (seeing the
changelog and associated diffs).  You could do this from within a squeak
image, of course, but that would require downloading each version you were
interested in.  Using a web interface to a Squeak image sitting on the
same machine as the version files is potentially a better use of
bandwidth, and allows for snappier browsing (particularly since that image
can cache the versions in memory of it wants to).

The other motive is that I wanted a Squeak image that could monitor the
repository directories and send out notification emails when new versions
appear.  ViewMC in fact grew up around the configuration interface for
this tool (and if you want to receive commit emails for any repository on
sqf.org, let me know and I'll add you).

> One thing I'm thinking of doing is saving the MC tools state in a file -
> instead of recreating the repositories/packages in each new image, save
> em and load them by default.

Yup, I've thought about that too.

> Another thing which I'd like but am not sure whats the best way to do is
> "open new versions" - suppose I'm working on Refactory, and I know there
> are two other people that have MC repositories where they occaisonally
> post patches. Would be nice to have an option that sees what their most
> recent work is, and for each package that isn't among my versions'
> ancestors, shows it in a list.
>
> I'm still not at the point where I really need it, because I don't have
> a high-turnaround collaborative project, but I do see the need for it.

Yeah, that would be nice.  One thing standing in the way of this is that
HTTP and FTP repositories have no good way of delivering just the
VersionInfo - you have to download the whole version.  So to find out
anything about ancestry you have to grab the entire .mcz files from all of
those repositories.  Maybe this isn't *that* big a deal, but it's a bit
annoying.

One thing to play with would be to save two files for each version - a
.mcz file that was complete, and a .mci file that just contained the
version info.  Or I wonder if it's possible to use the byte-range
extensions to HTTP to grab just the appropiate chunks of the ZIP file?