> At 01:26 PM 8/16/00 +0200, [EMAIL PROTECTED] wrote:
> >Hello,
> >
> >yes indeed, with setting case-fold-search to nil, then doing
> a Rescan, it
> >works.
> >But IMHO it�s not very good, if the whole stuff only works if
> >case-fold-search is nil.
> >There should be a solution regardless the value of this variable!
> >
>
> This requirement totally baffles me. Java is a case-sensitive
> language. So
> how can a case-insensitive parser possibly work? How, for
> example, can a
> case-insensitive Java parser distinguish between "Class" the class and
> "class" the modifier? I modified jde-parse to redefine
> case-fold-search as
> a local variable (i.e., in a let form) set to nil so that it
> would always
> do a case sensitive parse. Note that this change is entirely
> local to the
> parse. It does affect case-sensitive search in any other context. This
> small change immediately cleared up a bunch of parse errors
> in the files I
> tested. For example, the parser in JDE-2.2.2 parses the following
> declaration in java.lang.System class
>
> static native Class getCallerClass();
>
> as a class declaration. With case-sensitive parsing, it
> correctly parses it
> as a method declaration.
>
Maybe i you have misunderstood me. I don�t mean that the parser should work
case-insensitiv
but i should also work, if have set for my buffer case-fold-search to t because i want
search
in my buffer case-insensitive. IMHO it was the right way, to set case-fold-search
locally to nil
(with let...) in the parse-code, like you have described, so the parsing also works if
a user
want to do his searchings case-insensitive.
But in the current release 2.2.2 this is not the case, here parsing only works if the
buffer-local
value of case-fold-search is nil!
Makes this explanation things clearer?
Klaus
