Re: Another preproc portability problem with the new snarfer

[Matthias Koeppe]

Michael Livshin <address@hidden> writes:

> Matthias Koeppe <address@hidden> writes:
> 
> > Michael, did you receive my mail of last week?
> 
> yup, and replied to the list.  my mail to you bounces (I'm trying
> again now, but it probably won't work, 'cause all BigFoot addresses
> are apparently banned at your end).

I have only received one copy of this mail, the one from the
list.  (And one forwarded copy, thanks.)

> *strange*.
> 
> can you tell me how does the following command behave on your system?
> 
> echo 'argpos ( ( ( ( ( SCM ) ( ( ( const scm_bits_t * ) ( ( SCM_CELLPTR ) ( ( 
> ( scm_bits_t ) ( * args ) ) ) ) ) [ 0' | sed -e 's/^argpos.*[(\[].*//'
> 
> it is supposed to eat everything.  it does so on my Linux system at
> home and on the Solaris 2.5.1 and HP/UX systems at work.  is the sed
> incantation I use somehow non-portable?

It does eat everything here as well.  Maybe my last report was not
quite correct; I have now looked again at the snarfage fed into the
Guile script (after the sed scripts).  It contains the following:

========================================
(doc-block (
  
)
(
fname . "strerror"  
)
(
type . primitive  
)
(
location "error.c" . 156  
)
(
arglist . (err )  
)
(
argsig 1 0 0  
)
(
docstring

"Return the Unix error message corresponding to @var{err}, which\n"
"must be an integer value." 
)
(
))
(doc-check
argpos form 1 835 
)
(doc-check
argpos t . arg1 1 2823 
)
^^^^^^^^^^^^^^^^^^^^^^   
(doc-block (
  
)
(
fname . "apply:nconc2last"  
)
(
type . primitive  
)
(
location "eval.c" . 3275  
)
(
arglist . (lst )  
========================================

The Guile reader chokes on the form marked with carets.  

I hope this helps...

-- 
Matthias Köppe -- http://www.math.uni-magdeburg.de/~mkoeppe