Email List: Xaustin-review-lX
[All Lists]

Re: Defect in XCU Token Recognition

To: Mark Funkenhauser <yyyyyyyyyyyyy@xxxxxxxxxx>
Subject: Re: Defect in XCU Token Recognition
From: Miloslav Trmac <yyyy@xxxxxxxx>
Date: Tue, 29 Jul 2003 05:10:08 +0200
Cc: yyyyyyyyyyyyyyy@xxxxxxxxxxxxx
References: <200307282006.VAA07119@xxxxxx> <000401c35558$f95276c0$1e202b18@cr421871a>
Hello,
On Mon, Jul 28, 2003 at 06:38:31PM -0400, Mark Funkenhauser wrote:
> > There is at least one widely used
> > shell implementation (bash) that in fact does not honor the LC_CTYPE
> > category.
>
> Really ?  You think so ?
> That's not what the man page says.
> That's not what the source code says (I only have version 2.05b)
I was talking about honoring LC_CTYPE for <blank> interpretation in
token recognition.
* The whole bash source does not call isblank () or contain the "blank"
  character constant, except for interpreting bracket expressions in
  globs.
* parse.y, read_token_word () stops current word (in the "usual" case)
  when MBTEST(shellbreak (character)). shellbreak accesses sh_syntaxtab,
  which is initialized in mksyntax.c to mark characters from
  shell_break_chars, namely "()<>;&| \t\n".
Have I missed something?
  
> > Replacing "<blank>"
> > by "<space> or <tab>" promotes shell script portability and allows
> > faster shell implementations.
> I don't see how this wording change improves script portability.
I said promotes (encourages, if you want), not improves.
        Mirek

<Prev in Thread] Current Thread [Next in Thread>