Hi everybody, Sorry about the recent blip of automated bug-tracker traffic. I just manually fed a lot of recent changes to our automatic resolver
(Tip: the trick is to do darcs changes --xml -p 'resolve issue' > foo
and feed that in as the contents of an environment variable)
While I'm at it, I guess I could mention that I've made some progress on
the staging repo infrastructure.
This infrastructure does three things:
1. ensures that any patches I push pass tests
2. updates the bug-tracker automatically as people
resolve bugs (using the 'resolve issueXXXX' convention)
3. updates the API documentation on http://darcs.net/api-doc
Recent changes
--------------
- Scripts and settings now on
http://code.haskell.org/darcs/darcs-team
This almost means that all darcs hackers can run your own staging
repositories using the same set of scripts. Staging repositories
are useful because they avoid the embarrassment of pushing something
that breaks the build(bots). You'll want to comment out the part
that updates roundup, however, so that we don't get duplicates.
Either that or make it check that the bug isn't already resolved.
- Darcs/roundup integration should start working again. This means
we close bugs in a more timely fashion (provided I push patches to
staging on a regular basis)
- Patches don't actually get applied unless tests pass.
We use apply --test instead of darcs check --posthook; thanks, Petr
- Bug tracker is only informed of new patches if the apply
test passes (I hope! the whole thing now runs in a posthook;
I'm a bit disturbed that we seemed to get some updates last
night even though the tests failed...)
- API docs actually rebuilt
Previously, I was calling only cabal haddock --hoogle which meant not
updates to the HTML stuff, oops
Still broken...
---------------
The whole thing still depends on me remembering to push patches to my
staging repository. Perhaps #2 and #3 really ought to be offloaded to
a buildslave (or to the build master) so that we get that extra little
touch of automation.
I've also discovered a bad interaction http://bugs.darcs.net/issue1766
which was causing some bugs not to be updated.
Also, now that I'm using darcs apply --test on the staging repo, I'm
not getting very much output (probably a buffering issue).
Unfortunately, I never got around to working out why it was broken on
darcs.net <http://bugs.darcs.net/issue1542>, but at least we have
something that works via my staging repository. I think we shall just
have to satisfice there.
Finally, darcs.net really ought to be the one that runs haddock, but we
now seem to require haddock 2.5 for some reason, and darcs.net runs GHC
6.8.x (which I'm reluctant to change, because I think it keeps us
honest).
--
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9
signature.asc
Description: Digital signature
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
