Re: tetex-20011202 and bison-1.30g don't get along

2001-12-23 Thread Mike Castle

On Sat, Dec 22, 2001 at 03:47:09PM +0100, Akim Demaille wrote:
> Thanks for the report.  I believe the bug is (actually, are, there
> were two) fixed.  Could you give a try to
> 
> http://www.lrde.epita.fr/~akim/download/bison-1.30i.tar.gz
> http://www.lrde.epita.fr/~akim/download/bison-1.30i.tar.bz2

This version appears to work.

mrc
-- 
 Mike Castle  [EMAIL PROTECTED]  www.netcom.com/~dalgoda/
We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc



Re: tetex-20011202 and bison-1.30g don't get along

2001-12-22 Thread Akim Demaille

> "Mike" == Mike Castle <[EMAIL PROTECTED]> writes:

Mike> 1.30h fails with same error.

Thanks for the report.  I believe the bug is (actually, are, there
were two) fixed.  Could you give a try to

http://www.lrde.epita.fr/~akim/download/bison-1.30i.tar.gz
http://www.lrde.epita.fr/~akim/download/bison-1.30i.tar.bz2

?

TIA!




Re: tetex-20011202 and bison-1.30g don't get along

2001-12-19 Thread Akim Demaille

> "Mike" == Mike Castle <[EMAIL PROTECTED]> writes:

Mike> Whoops.  Deleted Akim's reply, so replying to myself to at least
Mike> keep it in the same thread.

Mike> On Mon, Dec 17, 2001 at 04:08:14PM -0800, Mike Castle wrote:
>> No obvious problems with bison-1.30e.  This is the first
>> compilation error I've encountered with bison-1.30g.

Mike> 1.30h fails with same error.

Mike> I generated a diff of the 1.30[eh] output, but even gzipped,
Mike> it's 12k.  More than I feel comfortable sending via email.  I'll
Mike> put it up on the web in a bit.  Hopefully I'll remember to write
Mike> the URL before I send this email.  (If the list owners want it
Mike> archived, they may to send it out anyway.)

Thanks, there is indeed a genuine problem somewhere.  The diffs are
big, but most of the diffs are irrelevant.  What is not is:

| @@ -471,7 +473,7 @@
|94, 0,   113, 0,   158, 0,   152,   154,94,94,
| 0,   245,   246,   241,   244,   243, 0, 0,   237,   300,
| 0,   271, 0,   291,   200, 0,   194,   198,   248,   260,
| - 139,   178,   180, 0, 0,   262,   125,94, 0,   124,
| + 139,   178,   180, 0, 0, 0,   125,94, 0,   124,
|   112,   115,   121,   122, 0,   119,   154,   156,   151,   157,
| 0,   173,   165,   169,   176, 0, 0,   276,   217,   219,
|   223,   225,   227,   229,   209,   211,   233,   213,   235,   215,

in the parser, and

| @@ -1708,7 +1713,6 @@
|  CONST_DEC_PART  ->  const_tok CONST_DEC_LIST .   (rule 33)
|  CONST_DEC_LIST  ->  CONST_DEC_LIST . CONST_DEC   (rule 35)
|  
| -undef_id_tok   reduce using rule 36 (@7)
|  $default   reduce using rule 33 (CONST_DEC_PART)
|  CONST_DEC  go to state 68
|  @7 go to state 57
| @@ -1852,7 +1856,6 @@
|  TYPE_DEC_PART  ->  type_tok TYPE_DEF_LIST .   (rule 82)
|  TYPE_DEF_LIST  ->  TYPE_DEF_LIST . TYPE_DEF   (rule 84)
|  
| -undef_id_tok   reduce using rule 85 (@26)
|  $default   reduce using rule 82 (TYPE_DEC_PART)
|  TYPE_DEF   go to state 78
|  @26go to state 72
| @@ -1944,9 +1947,6 @@
|  VAR_DEC_PART  ->  var_tok VAR_DEC_LIST .   (rule 126)
|  VAR_DEC_LIST  ->  VAR_DEC_LIST . VAR_DEC   (rule 128)
|  
| -undef_id_tok   reduce using rule 129 (@33)
| -var_id_tok reduce using rule 129 (@33)
| -field_id_tok   reduce using rule 129 (@33)
|  $default   reduce using rule 126 (VAR_DEC_PART)
|  VAR_DECgo to state 86
|  @33go to state 82
| @@ -3137,8 +3137,6 @@
|  VARIABLE  ->  var_id_tok . @51 VAR_DESIG_LIST   (rule 194)
|  VARIABLE  ->  var_id_tok .   (rule 195)
|  
| -'['reduce using rule 193 (@51)
| -'.'reduce using rule 193 (@51)
|  $default   reduce using rule 195 (VARIABLE)
|  @51go to state 252

etc. in the output.

I'll track this down, many thanks!




Re: tetex-20011202 and bison-1.30g don't get along

2001-12-18 Thread Mike Castle


Whoops.  Deleted Akim's reply, so replying to myself to at least keep it in
the same thread.

On Mon, Dec 17, 2001 at 04:08:14PM -0800, Mike Castle wrote:
> No obvious problems with bison-1.30e.  This is the first compilation error
> I've encountered with bison-1.30g.

1.30h fails with same error.

I generated a diff of the 1.30[eh] output, but even gzipped, it's 12k.
More than I feel comfortable sending via email.  I'll put it up on the web
in a bit.  Hopefully I'll remember to write the URL before I send this
email.  (If the list owners want it archived, they may to send it out
anyway.)

Out of curiosity, I tried byacc, and it worked fine.  :->

http://dalgoda.home.netcom.com/diff.gz

The teTeX source I'm trying to compile is at:

ftp://tug.ctan.org/tex-archive/systems/unix/teTeX-beta/teTeX-src-beta-20011202.tar.gz

Though it's a bit of a heavy-weight package to install.

Maybe we can get Thomas to generate a small example of the grammar and
input?

mrc
-- 
 Mike Castle  [EMAIL PROTECTED]  www.netcom.com/~dalgoda/
We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc



Re: tetex-20011202 and bison-1.30g don't get along

2001-12-18 Thread Akim Demaille

> "Mike" == Mike Castle <[EMAIL PROTECTED]> writes:

Mike> I'm not sure where there error occurs, so I'm sending to both
Mike> lists.  I'm not sure on subscription policies, so may need to
Mike> replies as necessary.

Mike> Joys of building every piece of cutting edge material I can lay
Mike> my hands on.  :->

:)

Mike> Oh, I guess I should first ask, can anyone else reproduce this,
Mike> or specific to my system?

No idea...

Could you try to diff the output of the two Bisons on the suspicious y
file?  Thanks!  Also, if there are useless rules in the grammar, then
I know 1.30g failed.  In this case, if there are indeed useless rules
in this grammar, please give a try to the forthcoming 1.30h.