Hi all,

This has been a pretty exciting week here, with my desktop PC dieing on
Tuesday. That turned the week upside-down rather, but it has had the
side-effect that since Friday I've had a shiny new 4-core machine, with
space for more GHC trees, and validate running in about 16 minutes.

The Mac also turned up on Friday, so I'm now able to test on OS X too.
It turned out to be about twice as large as I expected, but I've found a
home for it. Fortunately I can do everything over SSH, so no need for
comfortable physical access!

Interestingly, validating on my quad-core machine took 29 before I
upgraded python (to a version that supports parallel testsuite) and ghc6
(from 6.6.1 to 6.8.2). I'm not sure which of these made most of the
difference.


I've closed a few bugs this week:

#2537: Missing LICENSE file (which was causing install to fail) added
#2473: Exposed Data.Array.IO.Internals
#2541: Fixed install so that GHC can find utils like unlit
#2485: Fixed install so that ghci works

and also half-fixed #2542 (the "runghc Setup" problem), and I've done
the parts of #1338 (base breakup) that nobody objected to and moved the
bug to 6.12; we may be re-thinking #1338 though - more on that in a
later e-mail.

Given the plan to change the build system, we've decided not to spend
time fixing bindist for the current build system, and thus not to fix it
for the 6.10 release.


This week I was mostly working towards the 6.10 release, e.g. fixes for
the build system, to installing, and to making bindists, and working on
bugs. I've also been steering the full testsuite towards 0 failures, and
also fixing tests that could go wrong when running the testsuite in
parallel mode.

While working on other things in the base package, I found some of the
things I was fixing would have been caught by warnings, so I finished
fixing the warnings in base, and did unix while I was at it. Builds on
non-Windows now have warnings on for everything; hopefully we'll get
fewer silly little bugs introduced as a result!


Thanks
Ian

_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to