Mark Stosberg wrote:
On 2006-02-20, Tommy Pettersson <[EMAIL PROTECTED]> wrote:
  
This is what darcs 1.0.6 will look like, if noone finds anything
that needs to be fixed. I'm thinking mainly of build problems,
scripting and interfacing problems and undetected bugs
introduced by the fixes and new features.
    

FYI. Compiles and all tests pass on:

    FreeBSD 4.l0
    Perl 5.8.0
    ghc 6.2.2

    Mark

  

Same here, using the following configuration:

    Gentoo (almost everything latest stable)
    dev-lang/perl-5.8.7-r3
    dev-lang/ghc-bin-6.4.1
    sys-devel/gcc-3.4.4-r1

(with a re-emerged dev-tex/latex2html-2002.2.1_pre20041025, which got broken because of a Perl upgrade).

However, I have a couple of remarks.  Some of these might be already known, since it's been a time since I last built darcs myself.

First, I used http://darcs.net/darcs-1.0.6rc1.tar.gz for this test.  File Context.hs in this tar ball says that this contains the version with tag 1.0.6rc1 (created by Tommy).  However, this tag is not in the stable darcs repo, nor in darcs-unstable.  Therefore this tar ball is not reproducible except by Tommy.  Tommy, you do want to push this tag out to both stable darcs and darcs-unstable, don't you?

Second, when running "make check" to run all tests, the Perl test suite reported:

pull......................ok
        1/19 unexpectedly succeeded

If you want me to dig deeper into that, let me know.

Third, in "make check" the compilation of PatchTest.lhs resulted in the following

Warning: orphan instances:
  instance Test.QuickCheck.Arbitrary GHC.Base.Char = $fArbitraryChar
  instance Test.QuickCheck.Arbitrary FastPackedString.PackedString
    = $fArbitraryPackedString
  instance Test.QuickCheck.Arbitrary PatchCore.Patch = $fArbitraryPatch
  instance Test.QuickCheck.Arbitrary PatchInfo.PatchInfo = $fArbitraryPatchInfo
  instance GHC.Base.Eq PatchCore.Patch = $fEqPatch

I find it strange that this warning doesn't cause an error, in spite of the "-Werror" flag.  Is this perhaps an unavoidable warning?

Fourth, the last thing "make check" does is "ghc ... -o unit ..."; it does not actually *execute* this ./unit executable.  (Suggestion: Remove 'unit' from the dependencies of 'test' and 'check', and instead print a message there that says (after performing all the Perl tests) "do 'make test_unit' to perform a number of slower tests".)

Fifth, when running ./unit, a number of QuickCheck tests don't seem to actually test anything:

Checking patch and inverse both commute for subcommute simple_commute_conflict... Arguments exhausted after 0 tests.
Checking patch and inverse both commute for subcommute harder_commute_conflict... Arguments exhausted after 0 tests.
Checking patch and inverse both commute for subcommute repeated conflicts... Arguments exhausted after 0 tests.
Checking nontrivial commutes are correct for subcommute speedy_commute... Arguments exhausted after 0 tests.
Checking nontrivial commutes are correct for subcommute simple_commute_conflict... Arguments exhausted after 0 tests.
Checking nontrivial commutes are correct for subcommute harder_commute_conflict... Arguments exhausted after 0 tests.
Checking nontrivial commutes are correct for subcommute repeated conflicts... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute speedy_commute... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute simple_commute_conflict... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute harder_commute_conflict... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute simple conflicts... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute repeated conflicts... Arguments exhausted after 0 tests.
Checking inverses fail for subcommute force_commute... Arguments exhausted after 0 tests.

Hope all this helps.

Groetjes,
 <><
Marnix
_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users

Reply via email to