Re: [jamboost] gcc initialization revised

From: Michael Stevens (Michael.Stevens_at_[hidden])
Date: 2004-06-11 06:56:12

Volodya,

> I've just changed the gcc initialization according the the approach we've
> discussed recently.
>

Looks great! I try and see if I understand how you implemented this.

I think the only slight minus is the site-config case that David mentioned. 
But as long as the site-config only contains version qualified 'using' then 
the user-config can always add new version without error.

using gcc ;

> The most important new change is that "version" parameter should either be
> specified in all calls to "using", or not specified in any call.
>
> If the are no problems which this change, I'll look at applyting the same
> rule to other toolsets.
No problems as yet :-)
>
> Ah, and there's unrelated but nice change: there's a new
> 'errors.user-error' rule which is almost like 'errors.error', but shows
> only user modules in the backtrace. That's why the above messages don't
> mention toolset.using, gcc.init and common.check-init-parameters ;-)
The new error messages are very nice!

Michael