This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Simple RTL question (was: New cfg code)
- To: Jeffrey A Law <law at cygnus dot com>
- Subject: Re: Simple RTL question (was: New cfg code)
- From: Jamie Lokier <egcs at tantalophile dot demon dot co dot uk>
- Date: Tue, 3 Aug 1999 10:32:58 +0200
- Cc: Andi Kleen <ak at muc dot de>, Jerry Quinn <jquinn at nortelnetworks dot com>, egcs at egcs dot cygnus dot com
- References: <m3emhmhxlz.fsf@fred.muc.de> <3770.933655956@upchuck.cygnus.com>
Jeffrey A Law wrote:
> In message <m3emhmhxlz.fsf@fred.muc.de>you write:
> > Would it be possible to generate those parts of rtl.texi mechanically
> > from rtl.def (via a perl script or similar) ?
> It would be possible.
>
> However, in my experience, this does not actually work all that well. Just
> look at BFD which uses this kind of scheme.
Agree 100% on BFD.
My (entirely theoretical ;-) approach to this sort of thing is to
maintain docs and code together, and instead of mechanically generating
the entire docs, _verify_ the docs instead. So if the docs are missing
a field or function, the verifier script warns. Maybe it even inserts
the missing function with "not yet documented". Sometimes a good doc
string can be found with the thing to be documented, but sometimes it's
good to write something separately, that better fits a manual.
-- Jamie