[darcs-devel] summary of my recent spurt of patches, and request for suggestions

[David Roundy]

Hi all,

This is intended to be a summary of my activity over the weekend, for those
who are overwhelmed by the number of darcs sends I made--I kept thinking I
was done for the weekend.

I've been working hard on getting support for the new hashed inventory
format into good shape.  If you aren't familiar with the benefits of the
new format (which I've talked about with at least some of you in person),
suffice to say that I see it as a precursor to working out the new way of
dealing with conflicts.

I've gotten most of the tests to pass when run with hashed inventories, and
soon hope to get the rest under control.  In the process of generalizing to
a new repository format, I've been moving the code over to a more generic
framework (first introduced by Juliusz, but expanded by myself) which
abstracts out some of the details of the repository itself.

There are still several tests that fail, which I intend to track down:

TESTS FAILED!
        obliterate.sh
        partial.sh
        unrecord.sh

and 

Failed Test               Stat Wstat Total Fail  Failed  List of Failed
-------------------------------------------------------------------------------
optimize.pl                  2   512     4    2  50.00%  3-4
repodir.pl                   1   256    16    1   6.25%  10
set_scripts_executable.pl    1   256    25    1   4.00%  20

This is pretty small.  I don't know why the obliterate and unrecord
commands fail, but partial hashed inventory repositories are still on my
list to implement, so that's an understood failure, and optimize also
hasn't been updated to the new code.  Repair will also fail, but doesn't
seem to be in the tests--we should add a repair testcase or two.  I'm not
sure what causes the set_scripts_executable.pl error, but know where to
look.

Given these caveats, I'd appreciate it if interested persons could play
with hashed inventories, and see if they have any problems.  Patches to the
hashed_inventory.sh test would be most welcome, as I've mostly been running
tests with *all* hashed inventories, but am sure that some interesting bugs
will come out in the interaction between hashed inventory repositories and
old-fashioned repositories.

This is also the time to suggest changes to the actual format of the hashed
inventories.  It might be nice to add subdirectories to the patches/
directory, either by date or by the first couple characters of the hash
itself.  And I'm open to suggestions to make the format more parseable by
third party tools.  I just hacked this one together in... well quite a
short period of time, and since then have been focusing on the rest of the
code (which really doesn't depend on the inventory format in any
interesting way).

I guess that's all.  The term is over (which is why I've got a bit more
time and energy), but I've still got a final to give, and need to get
cracking on research (which was delayed by my class), so I'm not sure how
much darcs time I'll have in the upcoming weeks.  But it's likely to be
more than I had in the past couple of months.
-- 
David Roundy
Department of Physics
Oregon State University