On 3/10/2014 19:10, Francois Gouget wrote:
On Mon, 10 Mar 2014, Nikolay Sivov wrote:
On 3/10/2014 06:13, Jeremy White wrote:
Regarding Mac failures, it looks like this problem:
http://test.winehq.org/data/ecd24b5a874ead368c8f6e9d6981bb0e02472f9d/mac_fg-...
is about missing/old libxslt
What's interesting is that the macdrv tests do not run into this issue.
The difference is that for the macdrv tests I set DYLD_FALLBACK_LIBRARY_PATH="/opt/local/lib" which causes them to use what I believe to be the MacPorts libxslt.dylib library.
For the x11drv tests I set DYLD_FALLBACK_LIBRARY_PATH="/opt/X11/lib" because the libX11.dylib library does not (or did not) play well with XQuartz. As a result the X11 tests use /usr/lib/libxslt.dylib which is the library shipped with Snow Leopard.
I'll retry running the x11drv tests with /opt/local/lib.
However given that MacPorts is not part of Mac OS X that raises the question of what sort of system tweaks make sense in order to run the Wine conformance tests without error.
So it failed while running with system shipped version? I think we should use a version that we expect users to use, and I don't know which one is that. If system version is old enough to cause troubles then we should be using something more up-to-date from MacPorts. System lib is unlikely to be updated with system update, right? Especially for system that are not supported anymore (not sure if Snow Leopard still gets updates).
If it's decided to use system lib no matter what then test should be improved to skip in such cases.