Boost Testing :
From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2006-01-04 02:47:25
Reece Dunn writes:
> Martin Wille wrote:
>>Would it be possible to grab the first 32K and the last 32K of an error
>>message instead of the first 64K? I think it is quite likely that the
>>relevant portion is at the start or at the end of the output.
> Or if you change the BJam to XML converter to capture the output to a
> stringstream (output), then do:
> std::string line;
> while( std::getline( output, str ))
> std::cout << "<line>" << str << "</line>" << std::endl;
> this will solve the large output problem as it will be broken into
> line tags with text nodes that are small enough to be processed by
Something like that is definitely an option, but xsltproc not being
able to handle a large output is only part of the issue. Another part
is whether we want to that much of the data (almost 0.5 Gb in the
original case) to be processed, transfered over the net (potentially
making the runner to pay for the wasted bandwidth), and then loaded
into someone's browser.
The benefit of having it all in the reports is that the runner is
freed from requests to look up the corresponding info in her bjam log
in case the developer doesn't have access to the same platform.
Then of course we can make it a regression.py option.
-- Aleksey Gurtovoy MetaCommunications Engineering