[Mono-bugs] [Bug 542485] Fix for KeventWatcher

[bugzilla_noreply at novell.com]

http://bugzilla.novell.com/show_bug.cgi?id=542485

http://bugzilla.novell.com/show_bug.cgi?id=542485#c5


Romain Tartière <romain at blogreen.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |romain at blogreen.org

--- Comment #5 from Romain Tartière <romain at blogreen.org> 2010-05-06 07:00:20 UTC ---
For the ext[2] 64bit data structure, I have no idea about how to handle this. 
FreeBSD use a single structure for all architectures AFAICS. Maybe the code for
OSX 64 bit would need it's own watcher?


3.  I complete my answer: I did the cast because I realised that the code was
not working because it was basically doing:

requests [(int)fd] = ...

and then

.. = requests [(IntPtr)ev.ident]

Where both where the same "value" (e.g. 3) but in different types (and so the
lookup fails).  For this reason I though it was much more simple to promote the
only case we have an int to IntPtr and only have IntPtr in he hasktable.

Do you think it is safer to cast all IntPtr to int ?  There are not _many_
places where this should be changed (I actually have done it on my trunk
checkout... need to test but trunk is broken right now.  I'll test on the 26
branch).



I do not have test cases.  Do you have some test I could have a look to to
write some ?  I did not found reference to file watchers in mcs/test.  I think
we basically need a test that runs two threads, one actualy doing things and
the other waiting for these things to be notified.

Thanks!

-- 
Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.