Boost Testing :
From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-03-20 04:38:06
Martin Wille wrote:
> Vladimir Prus wrote:
>> Martin Wille wrote:
>>>after some reverse-engineering I was able to make regression.py not
>>>create another copy of the Boost tree. I also found out that
>>>regression.py apparently ignores the BOOST_BUILD_PATH environment
>>>variable and instead sets its own path (why?).
>> Good question. I don't know, apparently is happened this way before.
>> Would it be more convenient for you if this is fixed, by appending the
>> current directory to the value of BOOST_BUILD_PATH from the environment?
> That would be sufficient. However, I don't understand why that script
> should modify BOOST_BUILD_PATH at all if that variable is already set.
> (The same could be said about BOOST_ROOT)
No do I. Maybe, the idea is that custome toolsets can be added to this
directory where regression.py is run?
>>>However, I'm now stuck at this error message:
>>># Building "process_jam_log"
>> I'm a bit lost here. I've tried various permutations of version specified
>> in test-config.jam and version specified on the command line, the I could
>> only get the:
>> error: "3.2.3" is not a known subfeature value of <toolset>gcc
>> error message when the version is specified like this:
>> That is, with the dash instead of underscore.
>> Are you sure your test-config.jam has the right spelling? If yes, maybe
>> some other test-config.jam somehow is picked up? Can you go
>> to /boost/head-regression and run the bjam command:
>> /boost/head-regression/boost/tools/jam/src/bin/bjam --v2
>> -sBOOST_ROOT=/boost/head-regression/boost gcc-3.2.3_linux
>> --debug-configuration -n
> That command doesn't result in that error, and the correct
> test-config.jam gets picked up. Instead, I get:
> error: Unable to find file or target named
> error: '/boost/python//boost_python'
> error: referred from project at
> error: '../libs/python/test'
> (I didn't specify a Python path)
Yes, this is glitch, but an expected one. For now, the path to Python should
always be specified.
> I'll wait with further testing until the v1 tests that are currently
> running finish. I expect them to finish within the next 2 hours.
> I'll try to find out if something goes wrong in regression.py.