Re: [Boost-testing] Fix process_jam_long hangs by truncating message length

From: Beman Dawes (bdawes_at_[hidden])
Date: 2008-01-03 16:45:40

Nicola Musatti wrote:
> K. Noel Belcourt <kbelco <at> sandia.gov> writes:
> 
>> Hi,
>>
>> I've noticed that the Pgi compiler generates very long error messages  
>> (some in excess of 500k bytes).  Diagnostics this long cause  
>> process_jam_log to hang or get killed with a stack overflow message  
>> (probably caused by recursion in the split function).
>>
>> I'd like to propose we only process the first 8k bytes of an error  
>> message and throw away the rest.  Truncating the line length gets the  
>> Pgi test results reporting working reliably.  I've attached a simple  
>> patch to process_jam_log.cpp that I'd like to apply if there's no  
>> objections.  If people prefer, we could use a 16k limit instead of 8k.
> 
> Gasp! I had forgotten that I actually promised to look into this very problem!
> As you can see from this thread:
> http://thread.gmane.org/gmane.comp.lib.boost.testing/5129
> Beman himself appears to support the approach you suggest. 
> 
> Thanks for actually tackling this issue.

Ditto.

Go ahead and apply it. Let's try 8k to start with. Easy enough to change 
if anyone howls.

--Beman