[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: YYSTYPE_IS_TRIVIAL
From: |
Paul Eggert |
Subject: |
Re: YYSTYPE_IS_TRIVIAL |
Date: |
Fri, 3 May 2002 11:04:00 -0700 (PDT) |
> From: Akim Demaille <address@hidden>
> Date: 03 May 2002 17:24:59 +0200
>
> I would really like to know what problems you were experiencing.
It was confusing for me as well. I'll track it down if the problem
persists after gettext 0.10.3 is integrated.
> Muscle are really a dead simple thing: a hash table of key x value,
> and all the keys are output in M4 as m4_define(b4_key, its-value).
> Then the skeletons are put after this file containing the definitions,
> sent to M4, and post processed by scan-skel to handle quadrigraphs,
> #lines, splitting into several files etc.
"etc."? :-)
You're right, the muscles themselves are simple, but the overall
process isn't something I fully understand. For example, the current
CVS bison leaves behind droppings like /tmp/bsnW1ayDm if it fails on
Solaris, and I don't know why. I can investigate later (I'm a bit
pressed for time now).
I suspect that if I'm running into these problems that users will too,
and it may be necessary to have a brief mention in the manual of the
overall process, including what sort of phases and/or temporary files
that Bison uses. This could serve as a warmup for a larger part of
the manual that is less urgent, that would be mostly for maintainers
and tell them how to add new skeletons. (Easy for me to say, huh? :-)