Re: [Boost-build] [BBv2] PCH problems (was: removing propertiesfrom a property s

From: Reece Dunn (msclrhd_at_[hidden])
Date: 2005-11-23 11:02:52

Vladimir Prus wrote:
>On Wednesday 23 November 2005 11:57, Reece Dunn wrote:
> > Reece Dunn wrote:
> > > Q: Given a property set $(ps), how do I remove a property if I don't
> > > know it's value.
> > >
> > > e.g.
> > >
> > >     local newps = property-set.remove "<pch-header>" : $(ps) ;
> >
> > I have worked out howe to do this:
>
>The standard way is this:
>
>         local newps = [ property-set.create
>	                        [ property.change [ $(ps).raw ] : <pch-header> ] ] 
>;
>
>The "change" rule is used to change the value of feature to some other 
>value,
>but with empty value it just removes everything.

Cool!

> > with:
> >
> >     lib a : pch_a ... ;
> >     lib b : pch_b a ... ;
> >
> > we have:
> >
> >     PCH(A) --> LIB(A) -\
> >            PCH(B) --> LIB(B)
> >
> > where LIB(B) inherits both PCH(A) and PCH(B) properties. :(
>
>Yes, usage requirements are propagated infinitely. Is the above real-world
>case? Can you explain this to me? Should pch-related usage requirements
>*always* apply only to direct depenents? Are there any other corner cases?

This would be a real-world case. Especially for complicated projects that 
have many libraries all dependant on each other, all (or most) using PCH 
support.

>I think it's possible to make pch-related feature "propagated to direct
>dependents only".

That would solve this problem. Do you know how to do this?

> > The alternative that Alexey suggested of having:
> >
> >     lib a : ... : <pch>pch_a ;
> >     lib b : a ... : <pch>pch_b ;
> >
> > does not (I think) allow you to add a generator for the PCH file to
> > generate the *.pch file.
>
>What do you mean by "add a generator for the PCH file"?

I meant that with this alternative approach, it is not clear how the 
msvc.compile.pch rule+action (and the pch-generator) can be invoked to 
produce the PCH file.

>I think the biggest
>problem with this approach (assuming PCH rule is modified to add no usage
>requirements) is extracting header name from the target. I honestly don't
>know about right solution.

The propagate to direct descendants only will solve this issue and would be 
the right solution.

- Reece