🚀 go-pugleaf

Concerning: GNU Emacs 18.55.13 of Tue Nov 21 1989 on vega (berkeley-unix),
	    Run on a SparcStation 1 under SunOS 4.0.3.  Vega is a Sun-4/260.

I am using a number of complex regexps!  In the following narrowed-to
region:

"1989, p.453-65, 29 Refs, Country of Publ.: Netherlands"

a problem occurs at the blank between "Country" and "of".  The
following regexp fails:

"\\([0-9][0-9][0-9][0-9]\\),\\(  ?\\|\n\\)p\\.\\([---0-9+]+\\)\\([ \n]pp\\.\\)?
\\( vol\\.\\([^ ,]+\\)\\)?,\\(  ?\\|\n\\)[0-9]+\\(  ?\\|\n\\)Refs,
\\(  ?\\|\n\\)\\(.*,\\(  ?\\|\n\\)\\)*Country\\(  ?\\|\n\\)of\\(  ?\\|\n\\)
Publ\\.:\\(  ?\\|\n\\)\\(.*\\)$"

The physical linefeeds are there only for newsreader readability!  The
problem occurs in the "Country\\( ?\\|\n\\)" portion.  If the
alternating regexps after "Country" are replaced by "\\(  ?\\)", it
produces the following successful match-data:

(#<marker at 486 in *Backtrace*> #<marker at 540 in *Backtrace*>
#<marker at 486 in *Backtrace*> #<marker at 490 in *Backtrace*>
#<marker at 491 in *Backtrace*> #<marker at 492 in *Backtrace*>
#<marker at 494 in *Backtrace*> #<marker at 500 in *Backtrace*> nil
nil nil nil nil nil #<marker at 501 in *Backtrace*> #<marker at 502 in
*Backtrace*> #<marker at 504 in *Backtrace*> #<marker at 505 in
*Backtrace*> #<marker at 510 in *Backtrace*> #<marker at 511 in
*Backtrace*>)

The expression which matched is the following (\nl's are real linefeeds):

(looking-at "\\([0-9][0-9][0-9][0-9]\\),\\(  ?\\|
\\)p\\.\\([---0-9+]+\\)\\([ 
]pp\\.\\)?\\( vol\\.\\([^ ,]+\\)\\)?,\\(  ?\\|
\\)[0-9]+\\(  ?\\|
\\)Refs,\\(  ?\\|
\\)\\(.*,\\(  ?\\|
\\)\\)*Country\\(  ?\\)of\\(  ?\\)Publ\\.:\\(  ?\\)\\(.*\\)$")

Note that the length of the "(match-data)" is 20!!  The sub-regexp
matchs made after the ninth are *NOT* recorded!!  Does this have
anything to do with the TEN match constructs \&, \1, thru \9 ?  I
suppose so!

If so, it is either a GNU emacs error or a documentation error.  In
neither our GNU Emacs Manual (Fifth Edidion, Version 18) or our
preliminary GNU Emacs Lisp Manual (version 18, Sept. 1988), is there
mentioned a limit of TEN.  Only the "first 9 constructs" are accesible
as \<digit>.  In the documentation concerning the functions
match-data, match-beginning, match-end, etc., there is no mention that
the variable <count> must be less than 10!!

My diagnosis is that GNU Emacs ignores "\\("s and "\\)" over 9.
Unfortunately, if there follows an alternation "\\|" then it can not
backtrack correctly and therefor the match fails eventhough it is
syntactically correct.

Is there any other solution other than break-up the regexp?  Why is
there a limit on the number of allowed subexpressions?  The \<digit>
constructs could remain limited, but one could allow more flexibility
in Lisp.

As an aside, is gnu.emacs.bug gatewayed with the bug-gnu-emacs@prep.ai.mit.edu
mailing list?

--
Ralph P. Sobek			  Disclaimer: The above ruminations are my own.
ralph@laas.fr				   Addresses are ordered by importance.
ralph@laas.uucp, or ...!uunet!laas!ralph		
If all else fails, try:				      sobek@eclair.Berkeley.EDU
===============================================================================
Reliable software should kill people reliably! -Andy Mickel, Pascal News #13,78