[Top][All Lists]

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

Re: Two problems with the GLR parsers

From: Paul Eggert
Subject: Re: Two problems with the GLR parsers
Date: Fri, 15 Nov 2002 23:57:18 -0800 (PST)

> From: Akim Demaille <address@hidden>
> Date: Fri, 15 Nov 2002 16:37:23 +0100
> I'll try to extend the test ASAP.  I'll also try to fix the error
> handling in Yacc.  But I'm running low on time: I leave at the end of
> next week for a month :(

We really should get an official version of Bison out before then,
since the current official Bison version mishandles the Bash grammar.

The current CVS Bison has the following problems that I know of:

* The line numbers disagree in one GLR test case, which is what you're
  looking into as described above.

* GLR parsers fail badly on 32-bit Solaris SPARC when compiled with
  the Sun ONE Studio 7 C compiler see
  They work fine if I switch to 64-bit, or if I switch to GCC.  It's
  conceivable that it is a Forte bug, but more likely the bug is in

I guess I can try to look into the second problem but I'm afraid my
GLR-parser expertise is pretty limited.  Paul, if you want to take a
crack at it, do you have access to Sun ONE Studio 7 C?  It's also
called Forte 7 C.  You can get a free 60-day download from:

If we can't fix these two problems quickly, then perhaps the simplest
workaround is to change the data in the GLR test case to agree with
the current behavior (as flawed as it is).

reply via email to

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