Re: [Boost-build] [V2] Testing Tru64 toolset

From: Markus Schöpflin (markus.schoepflin_at_[hidden])
Date: 2006-02-10 11:14:43

Vladimir Prus wrote:

> On Thursday 09 February 2006 15:01, Markus Schöpflin wrote:
> 
>>> Well, I'm really confused what name is best ;-) Maybe, "hp-tru64", at
>>> least so that name is associated with the current vendor?
>> I just had a look at HP's web page, see
>> http://h30097.www3.hp.com/cplus/?jumpid=reg_R1002_USEN. They call it "HP
>> C++ Version 7.1 for Tru64 UNIX".
>>
>> So maybe hp-tru64 is a good choice, considering the name of the other
>> toolsets.
> 
> Ok, I'll rename it.

Thanks.

>>> What about renaming the feature, for V2, to "c++-abi". And one of the
>>> values would be "cxx-arm", or something? I think that <object-model>arm
>>> is not specific enough -- my first though that it selects object file
>>> format for the ARM processor ;-)
>> This makes sense. We would need cxx-arm, and cxx-ansi then, and the boost
>> default should be cxx-ansi.
> 
> I've impelemented some variation of that scheme. But default "-model ansi" is 
> used. And with "bjam --v2 c++abi=cxxarm", "-model arm" is added to command 
> line. That change is comitted.

Tested it with both examples and found it to work.

>> Good, I tested the second example and it works now. As soon as we have the
>> object-model back, I'll do a simple regression run.
> 
> That would be great! You might want to look at:
> 
>    http://zigzag.cs.msu.su/boost.build/wiki/TestingOnBoost
> 
> Note you'll need --v2 option both to process_jam_log and compiler_status. 
> Using regression.py is not officially supported yet -- I'm doing a test run 
> with it right now.

I started with the BBV2 regression tests as outlined in TestingNewToolset 
and I got a number of failures. Is this to be expected? If no, should I 
send you the log file of the test?

Markus