[ace-users] How well does the Proactor_Test work on Unix?

[Steve Huston]

Hi Craig,

> I am playing around with some patches that Matt Corey sent me
> a long time ago to add a kqueue() based implementation of the
> Proactor.  (Sorry for the delay Matt!!)

Cool.

> I was trying to test things out with the Proactor_Test.
> For some reason, it looks like this test deadlocks.  It doesn't
> seem to deadlock if I run the test in gdb.  The Proactor_Test also
> seems to deadlock if I run the real-time signal implementation
> under FreeBSD.  Does this test work with other Unixes?

Some yes, some no. Depends on the AIO implementation in the OS and
some varities of AIO completion handling work better than others on
most OSes. Check the scoreboard for details. Also, try running it with
1 thread to remove some complexity (./Proactor_Test -n 1)

> There have been a lot of changes recently in FreeBSD in the areas
> of threads, signals, and AIO, so I am using ACE as
> a vehicle for testing these features and learning how they work.

Ok.

> If the Proactor_Test is known to have issues, I don't want to let
> it hold up submission of patches for a kqueue() based Proactor.

The test itself is good, as far as I know (and I've worked with it a
lot).

-Steve

--
Steve Huston, Riverace Corporation
Would you like ACE to run great on your platform?
See http://www.riverace.com/sponsor.htm