[mapguide-internals] Mapguide 2.2 beta on Linux fails to setup locale from system

[Gabriele Monfardini]

> I took a quick look at the code and it should be setting the locale based on the LANG environment variable.  If LANG=en_US.UTF-8 then the MapGuide server should be using a UTF-8 character set.  The "C" setlocale will also look for LC_COLLATE, LC_CTYPE, LC_MESSAGES, LC_MONETARY, LC_NUMERIC, and LC_TIME before trying LANG.  If they are defined in the environment, it may be changing behaviour.

Thank you, Trevor,

I've opened a bug since system locale seems to be correct:

    $ locale
    LANG=en_US.UTF-8
    LC_CTYPE="en_US.UTF-8"
    LC_NUMERIC="en_US.UTF-8"
    LC_TIME="en_US.UTF-8"
    LC_COLLATE="en_US.UTF-8"
    LC_MONETARY="en_US.UTF-8"
    LC_MESSAGES="en_US.UTF-8"
    LC_PAPER="en_US.UTF-8"
    LC_NAME="en_US.UTF-8"
    LC_ADDRESS="en_US.UTF-8"
    LC_TELEPHONE="en_US.UTF-8"
    LC_MEASUREMENT="en_US.UTF-8"
    LC_IDENTIFICATION="en_US.UTF-8"
    LC_ALL=

I do not like also the fact that if an invalid character is found the
Error.log does not report anything even at the maximum verbosity, but
one apache child aborts.

I've always experienced random and not repeatable crashes without any
hint in any log (neither Trace.log, I've analyzed it extensively).
They are very annoying and tend to increase with load even with plenty
system resources.

The only hint I've found is a failing ACE assert that I've reported to
the mailing list some time ago, that usually is a consequence of a
failed interaction with the server (eg a failed setselection), but
very rarely lead to a crash.

In my humble opinion this kind of errors in multi-threaded
applications are impossible to debug without verbose logging.
Do you think it would be useful to open another bug against system
logging if I found anomalies that are note reported correctly in the
Error.log?