Fwd: C++ PATCH: PR 20599 (1/3)

[Mark Mitchell]

Mark Mitchell wrote:

> Just a quick note on this thread: I'm not ignoring it.  However, I may 
> not be able to comment on the what-should-we-do-about-C++0x for another 
> day or two, as I want to write something coherent, and I have other 
> commitments.

Here is my current thinking:

* From the point of view of FSF releases of GCC, it is not an objective 
for GCC to be an experimental implementation platform for programming 
languages.  Instead, the FSF releases are supposed to be compilers for 
people to use for production software development.  (I'm not claiming to 
speak for the FSF on this point; rather, it's my opinion, as a GCC SC 
member, about what the FSF's position ought to be.)

Therefore, we want to implement features because we think they will 
serve our users well over time, not because they will advance a 
standards process.  (In many cases, those goals will not be in conflict; 
but, in the cases where they are, our users should win.  And, of course, 
I'm all for people adding experimental features to GCC on branches, 
distributing those branches, etc.; my comments here are purely in the 
context of FSF releases.)

We should be sensitive to the fact that adding features and then 
changing them is going to make some of our users very unhappy.

* At the same time, it is in GCC's best interest to be the best compiler 
available, across all axes, and that includes supporting new programming 
language features.  If we can gain a competitive advantage by being the 
first to deliver a new feature, then we should.

* Therefore, building on the suggestions in this thread, my suggestion 
is that we:

(a) add C++0x features only with a command-line option (off by default, 
for now) so that users have to explicitly request the features,

(b) document that option in the manual as enabling experimental features 
and warning people that C++0x is subject to change, and that the GCC 
Gimplementation will track the actual standard, without regard for 
backwards compatibility with previous GCC releases,

(c) refrain from adding a feature until it is actually part of the WP, 
so that "feature oscillation" is minimized.

What do people think of that suggestion?

-- 
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713