My favourite way is (in the testsuite/ directory) to make accept (usually make TEST=<the tests I care about> accept), then inspect the diff with git diff (well, I use magit, but same difference, or whatever your favourite git diff inspector is).
That doesn’t help when I want to look at the output for 50 failed tests ... is this a gap in our current tooling?_______________________________________________On Tue, Nov 20, 2018 at 2:40 PM Ben Gamari <ben@well-typed.com> wrote:
On November 20, 2018 1:23:29 PM EST, Carter Schonwald <carter.schonwald@gmail.com> wrote:
>Happy to:
>
>What’s the best way to look at the Diffs / expected actual for 50 or so
>tests? So I can better triage these things as important or not going
>forward? For all I know it could be as simple as our passing diff
>stuff is
>white space significant. But I need some guidance on what files I
>should
>be looking at
>
>I can’t find any docs on how to poke at / read those on the wiki. Alp
>suggested just running a single test at a time, but that’s no ideal
>when
>it’s so many ...
>
>I ran that ghc build with clang from llvm 7 on high Sierra and no
>carter
>patches.
>
>I did cd testsuite ; make test THREADS=7
>
You can add TEST="T1234 T5678" to the "make test" command line. This is documented here [1]. Does this help?
Cheers,
- Ben
[1] https://ghc.haskell.org/trac/ghc/wiki/Building/RunningTests/Running#Commonlyusedoptions
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.
ghc-devs mailing list
ghc-devs@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs