From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-06-27 07:35:52
Rene Rivera wrote:
> [2003-06-26] Rene Rivera wrote:
> >[2003-06-26] Vladimir Prus wrote:
> >>Rene Rivera wrote:
> >>> [2003-06-25] Rene Rivera wrote:
> >>> >Stay tuned ;-)
> >>> OK, rewrite of the print.text rule is done. AFAIK no outside changes
> >>> are needed. But the use in response-file should not need to use the
> >>> "suffix" workaround, but it won't hurt either.
> >>Thanks! But, unfortunately, not everything is quite right. When I run
> >> test_all.py borland
> >>on windows, the 'generator_test.py' fails, because print.text invokes
> >>on empty string.
> >Hmm, shouldn't have happened :-( I'll fix ASAP...
> OK fixed... but I really hate the inability of the Windows ECHO to output a
> blank line :-(
Oh.. sure! And I don't understand my "cp" is called "copy" on windows ;-)
Thanks for fixing! All tests pass on borland now.
> In the process I was trying to run the generator_test, directly as I don't
> have Python on my Windows partition. And I decided I really didn't wan't to
> setup the user-config.jam file. I also miss the -sTOOLS=.. ability to
> init/specify the toolset. So I attempted to add a utility option "--tool
> <toolset> <options>*", but ran into one problem. The build system is not
> loaded at the time the options are parsed.
> So this brings up some questions:
> 1. Do we wan't to support options that need the build system loaded?
I don't know. The user-config.jam approach is fine with me, but I'm so used to
it that my opinion does not count.
> 2. If yes, where and/or how do we handle such a thing?
I think it's possible in a simple way. Allow an option module to contain a
rule, which tells if it should be executed after loading build system or not.
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