Re: GCC 3.4 Release Status (2003-12-16)

[Joe Buck <Joe dot Buck at synopsys dot COM>]

On Tue, Dec 16, 2003 at 11:32:15AM -0800, Geoff Keating wrote:

Mark Mitchell <mark@codesourcery.com> writes:
> > There are six open PCH bugs.  That's a new feature -- but it's one
> > that people are going to try to use, and, from their point of view,
> > problems with PCH that didn't occur without PCH are going to be
> > regressions. 

I disagree; people will expect that the first release of a major new
feature will have flaws that we do not catch.  Of course, we should give
the public as accurate an assessment of its quality as we can, but
it would be a mistake in my view to treat any open PCH bug so severely
(except in that it gives us permission to fix them all in 3.4.1 :-).

> > Alternatively, how would you feel about documenting PCH as
> > "experimental" so that people realize that this feature is not as
> > solid as we might like?

I think that we'll get a lot of questions, bug reports that aren't actually
bugs, etc. about it as people learn the conceptual model that it uses,
the restrictions, etc.  Not sure that this is a reason to call it
"experimental".  Call it "new", and mention any problems we think that
the user is likely to encounter.

(Whoops, sounds like a job for release notes.  Sigh).

> There are actually 8 PCH bugs.  None of these are serious enough to
> warrant delaying a release for.  They are all problems that are
> cosmetic, relate to constructs that don't usually appear in headers,
> or can be worked around in the build system.
> 
> I don't feel that PCH is not as solid as I would like.  It seems to
> have relatively few bugs compared to, for example, the C frontend.
> Apple has successfully used it to build an entire OS release.

Great, sounds like a strong test.

In your use at Apple, though, how many distinct pre-compiled headers
are there?  Pardon my ignorance, but are you just having all the
compilation units include one big .pch that has all of Carbon in it?