When we have filenames with non-ASCII characters in them, darcs record
and darcs whatsnew will report different filenames.

The good news is that this has been around since Darcs 2.3.1 perhaps
earlier (could somebody have darcs 1.0.9 on their Mac they could confirm
this with?)

Attached is the test I will push to the Darcs repository.

Since we have a test, the next we need to do is to investigate the cause
of the discrepancy.  Could be good to see what we're using to retrieve
these filenames in the first places, System.Directory?

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9

Attachments

• failing-issueXXXX-macosx-filepath.sh

With

$ uname -a
Darwin thorkil-naurs-mac-mini.local 9.8.0 Darwin Kernel Version 9.8.0:
Wed Jul 15 16:57:01 PDT 2009; root:xnu-1228.15.4~1/RELEASE_PPC Power
Macintosh
$ ./dist/build/darcs/darcs --version
2.4.4 (unknown)
$

I tried

> runghc Setup.lhs test tests/failing-issueXXXX-macosx-filepath.sh

with the attached failing-issueXXXX-macosx-filepath.sh, modified as follows:

> $ darcs w
> hunk ./tests/failing-issueXXXX-macosx-filepath.sh 9
> +PATH="/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:$PATH"
> +darcs --version
> $

The difference between the output of this test and the output of the
unmodified test is:

> $ diff
runghc_Setup_lhs_test_tests_failing_issueXXXX_macosx_filepath_sh_20100626_173757.out
runghc_Setup_lhs_test_tests_failing_issueXXXX_macosx_filepath_sh_20100626_174516.out
> 42,45d41
> < PATH="/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:$PATH"
> < +
PATH=/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/Volumes/tn18_HD_1/Users/thorkilnaur/tn/darcsDarcsRepository/darcs.net_stable-for-misc-work-20081118_2206/darcs.net/dist/build/darcs:/Users/thorkilnaur/tn/install/ghc-6.10.4/bin:/Users/thorkilnaur/tn/install/hscolour-1.10.1/bin:/Users/thorkilnaur/tn/install/darcs-2.4/bin:/Users/thorkilnaur/tn/install/maybench-0.2.4.1/bin:/Users/thorkilnaur/tn/install/haddock-0.8/bin:/Users/thorkilnaur/tn/install/happy-1.17/bin:/Users/thorkilnaur/tn/install/alex-2.2/bin:/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/opt/local/bin:/opt/local/sbin:/Users/thorkilnaur/tn/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin
> < darcs --version
> < + darcs --version
> 61d56
> < 1.0.9 (release)
> 124,128d118
> < PATH="/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:$PATH"
> < 1.0.9 (release)
> < +
PATH=/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/Volumes/tn18_HD_1/Users/thorkilnaur/tn/darcsDarcsRepository/darcs.net_stable-for-misc-work-20081118_2206/darcs.net/dist/build/darcs:/Users/thorkilnaur/tn/install/ghc-6.10.4/bin:/Users/thorkilnaur/tn/install/hscolour-1.10.1/bin:/Users/thorkilnaur/tn/install/darcs-2.4/bin:/Users/thorkilnaur/tn/install/maybench-0.2.4.1/bin:/Users/thorkilnaur/tn/install/haddock-0.8/bin:/Users/thorkilnaur/tn/install/happy-1.17/bin:/Users/thorkilnaur/tn/install/alex-2.2/bin:/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/opt/local/bin:/opt/local/sbin:/Users/thorkilnaur/tn/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin
> < darcs --version
> < + darcs --version
> 206,210d195
> < PATH="/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:$PATH"
> < 1.0.9 (release)
> < +
PATH=/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/Volumes/tn18_HD_1/Users/thorkilnaur/tn/darcsDarcsRepository/darcs.net_stable-for-misc-work-20081118_2206/darcs.net/dist/build/darcs:/Users/thorkilnaur/tn/install/ghc-6.10.4/bin:/Users/thorkilnaur/tn/install/hscolour-1.10.1/bin:/Users/thorkilnaur/tn/install/darcs-2.4/bin:/Users/thorkilnaur/tn/install/maybench-0.2.4.1/bin:/Users/thorkilnaur/tn/install/haddock-0.8/bin:/Users/thorkilnaur/tn/install/happy-1.17/bin:/Users/thorkilnaur/tn/install/alex-2.2/bin:/Users/thorkilnaur/tn/install/darcs-1.0.9/bin:/opt/local/bin:/opt/local/sbin:/Users/thorkilnaur/tn/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin
> < darcs --version
> < + darcs --version
> $ 

I would conclude that this problem existed for darcs-1.0.9 as well.

And, by the way: This test also fails with:

> $ uname -a
> Linux tn24 2.6.32-22-generic #36-Ubuntu SMP Thu Jun 3 22:02:19 UTC
2010 i686 GNU/Linux
> $ ./dist/build/darcs/darcs --version
> 2.4 (+ 142 patches)
> $

Best regards
Thorkil

On Sat, Jun 26, 2010 at 16:03:28 +0000, Thorkil Naur wrote:
> I would conclude that this problem existed for darcs-1.0.9 as well.

Many thanks!  That's very comforting.  It's nice to see that we're
starting into the old-bugs territory and not just cope with our
regressions.

> And, by the way: This test also fails with:
> 
> > $ uname -a
> > Linux tn24 2.6.32-22-generic #36-Ubuntu SMP Thu Jun 3 22:02:19 UTC
> 2010 i686 GNU/Linux

OH! Good thinking...  huh! I wonder why the issue1763 test works on
Linux but not on OS X then.  Perhaps issue1763 sets the locale just
so...

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9

I have a note from myself asking me what "locale" prints on my Mac:

LANG="en_US.UTF-8"
LC_COLLATE="en_US.UTF-8"
LC_CTYPE="en_US.UTF-8"
LC_MESSAGES="en_US.UTF-8"
LC_MONETARY="en_US.UTF-8"
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_ALL="en_US.UTF-8"

It was originally for issue1763, but some recent past version of myself
said issue1880 may be more appropriate.

This should be fixed by now. We have lots of tests for this and they 
all succeed (nowadays).