Re: Shake fails test with GHC 7.10 RC3

Yes, this was what I was afraid of; an introduced regression in the RC 2-to-3 period, which IMO makes the bug a bit more serious. Also, this bug in particular looks like it manifests in a fairly nasty way, and having it catch people for the final release seems a bit worrisome. I also just remembered that the Haddock documentation for 7.10.1 is still broken; I've been tracking the bug today, but I need to sync up with the Haddock maintainers to discuss more. I really think this one needs to be fixed. So I also suggest we postpone for a few days to fix these two things, if possible. On Sat, Mar 21, 2015 at 6:55 AM, Neil Mitchell wrote:

Interesting result. I've just attached a much smaller test case to the ticket - it would be good to see if that fails or passes with RC2, or breaks at the same point.

The fact that there does seem to be a regression between RC2 and RC3 (either a regression itself, or making a previous bug easier to hit) makes me think that a release might not be such a great idea.

Thanks, Neil

On Sat, Mar 21, 2015 at 11:35 AM, Herbert Valerio Riedel wrote:

...

On 2015-03-21 at 08:21:20 +0100, Herbert Valerio Riedel wrote:

...

On 2015-03-21 at 07:56:32 +0100, Neil Mitchell wrote:

[...]

...

3) I tested with GHC RC1 and GHC RC2, both of which were fine. The fact no one else hit this with RC2 might just be because its a very recent regression.

We -- and by that I don't mean myself... :) -- could git-bisect between RC2 and RC3 here (semi-)automatically (i.e. maybe unattended if it's scriptable) if your test-case (even if it's not minimal) reliably triggers the bug...

I scripted up a test and git-bisected between RC2 and RC3, and the following commit is the one where `shake-test oracle test` starts failing

http://git.haskell.org/ghc.git/commitdiff/6f46fe15af397d448438c6b93babcdd68d...

...which sadly is a rather large patch :-/

-- Regards, Austin Seipp, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/