Re: [Boost-testing] process_jam_log fails with bad_alloc exception

From: Beman Dawes (bdawes_at_[hidden])
Date: 2007-11-06 17:16:31

Nicola Musatti wrote:
> Doug Gregor wrote:
>> On Oct 31, 2007, at 2:37 AM, Nicola Musatti wrote:
>>> By running pjl with the --echo option I discovered that the problem is
>>> caused by the graph/dominator_tree_test outputting enough  
>>> repetitions of
>>>    2147483647 to fill an over 400,000 character long line!
>>>
>>> I'm not sure whether this is supposed to be normal behaviour or it  
>>> just
>>> happens because the test was timed out.
>> Ouch! I see the bug in Trac and will ping the author of that  
>> particular algorithm. If we don't get a fix soon, I'll dig into it or  
>> yank the test until we have a solution.
> 
> Unfortunately this problem was still present as of last night, CET time. 
> I'm currently trying to run the Borland debugger on 
> dominator_tree_test.cpp, but it just crashed on me. I'll try and be more 
> careful next time around.
> 
> In my opinion we should tackle this problem from two sides: trying to 
> avoid such an unwieldy output from this test, but also ensuring that pjl 
>   truncates overly long line to a reasonable size. I would venture that 
> 10KB would be a reasonable size, given how certain compilers fully 
> expand some template definitions.

Makes sense to me. Care to submit a process_jam_log.cpp patch?

--Beman