On Feb 05 2003 19:00, Brane Čibej <brane@xbc.nu> wrote:
> Dr.Stefan.Dalibor@bfa.de wrote:
>
> >I confirmed this under Cygwin - if the following patch is applied to
> >commit_tests.py, the test passes (I'm just not sure which would be
> >the correct substitution character for the angle bracket under
> >Cygwin):
> >
> >***
> /tmp/subversion-r4503/subversion/tests/clients/cmdline/commit_tests.py-orig
> Wed Jan 22 08:16:09 2003
> >---
> /tmp/subversion-r4503/subversion/tests/clients/cmdline/commit_tests.py
> Wed Feb 5 09:51:59 2003
> >***************
> >*** 1040,1045 ****
> >--- 1040,1053 ----
> > if svntest.main.windows:
> > angle_name = '$angle$'
> > nasty_name = '#![]{}()$$%'
> >+ elif sys.platform == "cygwin":
> >
> Why don't you make this a single test?
>
> if svntest.main.windows or sys.platfor == "cygwin":
You can't test for <> under M$ (Win32API limitation), you can't
test for $ under Unix (at some point, the shell kicks in and replaces
$$ by its PID), and you can't test for both of them under Cygwin
(Unix shell on Win32API).
This patch was just a guess - I don't fully understand which relevance
each of the precise `nasty' characters has for the test. Probably the
real fix would be to suppress the expansion of $ under Unix (I played
around with inserting backslashes before the $, but that didn't cut
it).
Then the if test could just stay as distributed with r4503... in the
end, it's meant to work around platform inherent file system
restrictions, not interpreter quirks, isn't it?
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Feb 6 18:42:16 2003