Re: [PCH] Automatic PCH creation model

[Karel Gardas <kgardas at objectsecurity dot com>]

On Fri, 29 Aug 2003, Neil Booth wrote:

> Karel Gardas wrote:-
>
> > I would like to ask if anybody is working on automatic PCH creation - like
> > it is working in Comeau's C/C++ compiler and IIRC in Borland's C++
> > compiler. e.g. let's have some options which directs usage of pch -
> > ``--pch'' and ``--pch_dir'', then for example
> >
> > c++ <params> --pch -c a.cc
> >
> > will create a.pch in the same dir like it creates a.o. Also if the next
> > command is:
> >
> > c++ <params> --pch -c b.cc
> >
> > and the a.pch is suitable for reuse in b.cc compilation, then it is used.
> > If it is not, then b.pch is created. In the process of source unification,
> > there are some #pragmas which might be used to directs pch processing,
> > like #pragma hdrstop and #pragma no_pch. In case of #pragma no_pch, pch
> > usage will be disabled for this file and in case of #pragma hdrstop, pch
> > processing - i.e. the set of include files till the #pragma hdrstop will
> > be used to find the best matching pch file. More about this scenario is
> > written in como documentation on
> > http://www.comeaucomputing.com/4.0/docs/userman/pch.html.
>
> Indeed.  I much prefer this model (it's specified if nothing else!) and
> said so in Jan 2001.  Sadly the PCH implementation went down a different
> road.

I also prefer this model and that was the reason for asking here on
gcc-devel. Anyway as you see como also provides manual PCH creation model
and so the question is if it is possible to write pch automatic creation
model for gcc too - i.e. if it is worth the time.

Also there is the compiler server project and if it provides some server
automatic startup support, then it might be used instead of gcc pch.

Problem: I can only volunteer as a C++ tester using MICO project for any
compilation time meassurements.

Cheers,

Karel
--
Karel Gardas                  kgardas@objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com