This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Configuration/Makefile cleanups
>>>>> "Marc" == Marc Espie <espie@quatramaran.ens.fr> writes:
Marc> On Fri, Nov 24, 2000 at 04:31:14PM -0800, Mark Mitchell
Marc> wrote:
>> One of the tools we require is bison. It's nice for us to have
>> a standard parser generator (same features/bugs everywhere),
>> and bison the GNU parser generator, so that's what we use.
Marc> Considering that byacc has the same features, I challenge
Marc> that conclusion.
Marc> Being able to use interchangeable tools is a proof of
Marc> robustness, as you depend only on the published interface,
Marc> not the specific set of bugs of the tool you're using.
That's true.
There's always going to be a tradeoff between that benefit, and the
associated cost of having to deal with multiple tools. In some cases
(like using a somewhat broken K&R C compiler to build GCC), the cost
is high -- but the benefit is high, too, since otherwise GCC couldn't
be built at all on lots of systems.
In this case, neither the cost, nor the benefit is very high. In my
opinion, the cost is, however, greater than the benefit. Therefore,
it's my opinion that we needn't have Makefile gunk to support byacc.
I don't think that cost is worth the marginal benefits of a) saving
BSD people the trouble of building Bison, or b) finding latent bugs in
our parsers that are somehow not exposed by Bison.
If there's consensus otherwise, or the Steering Committee decides that
it wants to tackle this issue, that's fine by me.
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com