Known Issues
This is a list of known issues with the current release of BRAHMS, along with advice on how to deal with them if you encounter them.

BRAHMS 0.7.2

Version

Issue

Advice

Status

0.7.2
Abort/Exception when MaxThreadCount is parsed
A bug will cause an abort (windows) or an exception (linux) when MaxThreadCount is set to "unlimited". "unlimited" is indicated either by setting the value to "0" (zero) or to "" (the empty string). Confirm you have this problem by passing "--detail-5" to the executable; the failure should occur immediately after a message about "MaxThreadCount". Workaround is to set the value of MaxThreadCount to "999999" (i.e. practically unlimited).
Workaround
Set the value of MaxThreadCount to "999999" (i.e. practically unlimited).
 Fixed in HEAD
0.7.2
Matlab/Python outputs corrupted when process has more than one output
A bug will corrupt output of Matlab/Python processes when those processes create more than output. Alternatively, on some installations, an abort may result.
Patch
Patch 2094 (see download) fixes this issue.
 Fixed in HEAD
0.7.2
Incorrect Base Sample Rate displayed
A bug will cause the run-time GUI to display Base Sample Rate incorrectly if higher than 99999Hz. No other effects (computations are performed correctly).
Workaround
None required.
 Fixed in HEAD
R2230
MPI execution does not pick up user-level Execution Parameters (Windows)
Since individual processes run under MPI as "LocalService", rather than the user that launched them, the user's Execution Parameters are not picked up.
Workaround
Set any Execution Parameters at the machine-level (see "brahms_manager" under Matlab).
 Not Fixed
R2233
Following a hung thread on Linux, a segmentation fault may occur
pthreads doesn't mix well with C++ exceptions, leaving us unable to kill hung threads on linux. All we can do, afaik, is unload the modules under them. This, it seems, results in an abort (which i think makes sense). See developer note "pthread_kill".
Workaround
Whilst this is not an elegant state of affairs, there's no real hit, since the abort happens after the thread has been abandoned by the framework anyway. Ignore the message, fix the hang, and get on with your life.
 Not Fixed
0.7.2
"Invalid return value for stat" when running under Matlab
If running under Matlab, and running a system that includes a Matlab process, the Matlab engine may fail to start in an ugly way. The message you will see will be something like:
>> brahms_test language m
Severe:
I18N Runtime Error:
	Invalid return value for stat detected while processing $(MATLAB)/bin/$(ARCH).
	Hint: Notify MathWorks support.
The program '[16268] : Native' has exited with code 2 (0x2).
and may be followed by an abort. I have literally no idea what this means, and Google hasn't helped, but it seems to only happen if you are running in the Matlab "bin" folder.
Workaround
Don't run "brahms" (or, similarly, "brahms_test") when the current directory (pwd) is the Matlab bin folder (e.g. "/usr/local/matlab/bin").
 Not Fixed - this may be a Matlab bug