From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-07-20 08:43:42
At 08:42 PM 7/18/2002, David Abrahams wrote:
>> >It won't help because all those toolsets use the same actions. Your
>> >analysis should not rely on the action name for anything.
>> The action name is the only way it knows what it is looking at.
>Why don't you just look at the name of the target it's building? That
>always has the correct toolset name in it.
That's what I do for toolset name. But I also need to capture
compiler/linker/run messages and other information. That requires looking
at the action name to be able to distinguish which of the
compiler/linker/run messages follow.
>> Building in more action names isn't a problem; it is just messy.
>"just messy?" You clearly don't have to maintain the build system.
I meant "just messy?" for the analysis program, not for the build system.
>> >P.S. I've often said that trying to postprocess the results of the
>> >process with an external tool is not likely to work well, and I think
>> >are running into many of the problems I was thinking of.
>> Actually, I was having a lot of trouble with the build residue.
>That was part of what I thought would be unreliable.
>> Once I
>> switched to using the log analysis to update a permanent .xml file,
>> everything start working much better.
>I can see how that would help. I didn't realize you were doing that.
Bill Kempf, supported by several other people, suggested I do that. It
turned out to be much easier than I thought, has solved several problems,
seems much more robust, and is supporting new features like capturing
warning messages. I'll post an example soon.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk