From: Reece Dunn (msclrhd_at_[hidden])
Date: 2005-10-27 15:54:55
Zbynek Winkler wrote:
> :) I think we should start reading each others emails. IMO what has
> happened here is that each one of us saw the light at various points of
> time. So I guess that we all are proposing
> actions compile.c++
> mv @(INCLUDES) $(<).rsp
> cl $(<).rsp ....
> where the @() syntax takes the variable, outputs its contents to a
> temporary file and returns its filename instead.
Yes! Now... how to port the functionality to bjam from Matt's code.
> And just for the record - I would *really* love to see the
> hashed-command-line mechanism to rebuild things. I just see the problem
> of limited command line length and the inclusion of the command to the
> dependency graph as orthogonal.
You would need to include the contents of the response file in the hash
as well. Hashing command line + RSP would be enough to detect changes,
but where to store these hashes (and avoid the current RSP file problem).
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