Re: [Boost-testing] BGDX failures on gcc-4_0-darwin

From: Joaquin M Lopez Munoz (joaquin_at_[hidden])
Date: 2005-09-03 14:48:07

Beman Dawes <bdawes <at> acm.org> writes:
> 
> Even though the same gcc build is now being used for BGDX and dmcgrath, some 
> results still differ. Multi-index tests, for example. I don't have a clue as 
> to why the results differ.
> 
> The multi-index tests hung. I had to manually kill the process each time.
> 

Hello Beman,

A few days ago I checked in some synchronization code in multi_index
(with lightweight_mutexes) to eliminate potential problems
when in multithreading mode. Maybe this can be the culprit?
The synchronization code is extremely simple (http://tinyurl.com/9jzby),
and moreover the tests do not launch any thread, but who knows.

1. BGDX is running in multithreading mode (BOOST_HAS_THREADS
defined.) Could you try switching to singlethreaded and see
if the tests no longer hang?

2. I cannot know whether dmcgrath is running in multi- or
singlethreaded mode (config_info test seems to fails,
see http://tinyurl.com/aduy4.) It'd be useful to have this
info: if dmcgrath environment does not define BOOST_HAS_THREADS,
then we have an actual difference between both platforms.

HTH a little,

Joaquín M López Muñoz
Telefónica, Investigación y Desarrollo