[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: FYI: Both: Gr. Reduction
From: |
Akim Demaille |
Subject: |
Re: FYI: Both: Gr. Reduction |
Date: |
28 Nov 2001 15:50:51 +0100 |
User-agent: |
Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Artificial Intelligence) |
| It looks like I broke something recently :*(. 1.30 dies, including
| with the patch below, when producing the verbose report (while reading
| tags[-1]). Also, be frightened, be very frightened:
|
| | ~/src/bison-1.29 % cd tests/testsuite.dir/17
| | tests/testsuite.dir/17 % cat input.y
| | %verbose
| | %output="input.c"
| |
| | %nterm useless1
| | %nterm useless2
| | %nterm useless3
| | %nterm useless4
| | %nterm useless5
| | %nterm useless6
| | %nterm useless7
| | %nterm useless8
| | %nterm useless9
| |
| | %token useful
| | %%
| | exp: useful;
| | tests/testsuite.dir/17 % cat input.output
| | Useless nonterminals:
| |
| | exp
| |
| |
| |
|
| Ahem...
My first guess is that this is the result of my separating generation
from output: I believe that this output used to be produced *before*
the reductions was actually performed. Now it is certainly done
*after* the reduction was performed. Therefore, it looks at the slots
where there used to be useless things, which have been removed since
then.
So it is a ``good'' bug: it is the results of something we want
(separation between input/process/output), but it requires some more
work.
I'll try to give it a stab this evening.