bug-bison
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: bison-2.1 compilation warnings


From: Frank Heckenbach
Subject: Re: bison-2.1 compilation warnings
Date: Fri, 12 May 2006 22:18:24 +0200
User-agent: semail 20060101

Akim Demaille wrote:

> Le 11 mai 06 à 09:25, Paul Eggert a écrit :
> 
> > I vaguely recall mentioning this to W.L. Estes (the flex maintainer, I
> > guess) a while ago, and he responded that he didn't care about
> > generating code for pre-C99 compilers.  A bit extreme, in my opinion.
> 
> Quite extreme in mine :)
> 
> > The flex incompatibility remains unfixed in flex 2.5.33.

It's worse. I recently had a conversation with John Millaway, the
current flex maintainer. It was (apart from a few minor bugs) about
some flex functions output without prototypes which would cause
warnings with "gcc -Wmissing-prototypes". Strangely, the prototypes
in question are actually present in the flex skeleton, just not
output on the same conditions as the functions, I suspect due to a
mismatch of conditionals (of which there are a lot indeed) between
the prototypes and the functions sections.

But instead of checking and fixing the conditionals, he kept talking
of "bogus warnings" and "a GCC problem" (even though I explained to
him that I intentionally chose the GCC option which does what it
says), and "ridiculous", and finally, I quote: "Style policing and
best-practice mumbo jumbo are for the birds."

He went on how complex the flex output is, with all those options
and modes, apparently implying that it's impossible to even care
about compiler warnings in such a program. Of course, every
bug-bison reader knows better. :-)

So I guess, with this attitude of his, things won't improve on the
flex side, probably even get worse, as long as he's the maintainer.

But BTW, he also mentioned that he didn't really want to be the flex
maintainer, and just picked it up because the previous maintainers
(Vern Paxson and W.L. Estes) couldn't do it anymore, and noone else
was found.

So perhaps that's an opportunity. If there's someone here who'd like
to work on such a program and has a little time to spare (I don't,
unfortunately), they might relieve John Millaway from an unpleasant
job and help fix such problems on the flex side, so we won't have to
work around them forever (which goes for Bison, as in this thread,
and for my and other people's flex-using programs as well).

Frank

-- 
Frank Heckenbach, address@hidden
http://fjf.gnu.de/
GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)




reply via email to

[Prev in Thread] Current Thread [Next in Thread]