[9fans] grep (or perhaps regex(6)) question

[9fans] grep (or perhaps regex(6)) question

[dexen deVries]

on plan9port:
% echo foobar | 9 grep '$JUNK'
 -> foobar

that is, p9p grep matches any line with regular expression in form of
$ANY_REGEX

why is it so?

-- 
dexen deVries

[[[↓][→]]]

For example, if the first thing in the file is:
   <?kzy irefvba="1.0" rapbqvat="ebg13"?>
an XML parser will recognize that the document is stored in the traditional ROT13 encoding.

(( Joe English, http://www.flightlab.com/~joe/sgml/faq-not.txt ))

Re: [9fans] grep (or perhaps regex(6)) question

[Rudolf Sykora]

> that is, p9p grep matches any line with regular expression in form of
> $ANY_REGEX
>
> why is it so?

Why (the very mechanism) is one question. Whether it should or should
not be so is another. Is this subtlety important? (I.e. can't it be
implementation dependent as the correct behaviour seems undefined [to
me]?)

Thanks
Ruda

Re: [9fans] grep (or perhaps regex(6)) question

[erik quanstrom]

On Tue Jul  5 03:57:02 EDT 2011, dexen.devries@gmail.com wrote:
> on plan9port:
> % echo foobar | 9 grep '$JUNK'
>  -> foobar
>
> that is, p9p grep matches any line with regular expression in form of
> $ANY_REGEX
>
> why is it so?

this is a bug.  see the grammar in regexp(6).
plan 9 grep does not have this problem.

- erik

Re: [9fans] grep (or perhaps regex(6)) question

[Rudolf Sykora]

>> that is, p9p grep matches any line with regular expression in form of
>> $ANY_REGEX
>>
>> why is it so?
>
> this is a bug.  see the grammar in regexp(6).
> plan 9 grep does not have this problem.
>
> - erik

I am sorry, but I don't understand.
What do you mean is wrong, Erik?

Thanks
Ruda

Re: [9fans] grep (or perhaps regex(6)) question

[erik quanstrom]

On Tue Jul  5 08:42:16 EDT 2011, rudolf.sykora@gmail.com wrote:
> >> that is, p9p grep matches any line with regular expression in form of
> >> $ANY_REGEX
> >>
> >> why is it so?
> >
> > this is a bug.  see the grammar in regexp(6).
> > plan 9 grep does not have this problem.
> >
> > - erik
>
> I am sorry, but I don't understand.
> What do you mean is wrong, Erik?

there should be no match to the re $junk.
it's impossible for there to be text after then
end of line.  plan 9 does this correctly,
my version of p9p (which may differ by
being out-of-date) does not.

- erik

Re: [9fans] grep (or perhaps regex(6)) question

[Rudolf Sykora]

> there should be no match to the re $junk.
> it's impossible for there to be text after then
> end of line.

1) from the regexp grammer, $junk is a valid regexp
2) grep checks input line-by-line
3) the answer to a question whether a *line* complies with '$whatever'
is --- I believe --- undefined. (It definitely should match '$';
however comparing with '$whatever' gets us outside the definitions).

Ruda

Re: [9fans] grep (or perhaps regex(6)) question

[erik quanstrom]

On Tue Jul  5 10:02:56 EDT 2011, rudolf.sykora@gmail.com wrote:
> > there should be no match to the re $junk.
> > it's impossible for there to be text after then
> > end of line.
>
> 1) from the regexp grammer, $junk is a valid regexp

i believe you've read incorrectly.  but if you think it parses,
what production produces e3 en?  but that wasn't actually my point.

> 2) grep checks input line-by-line
> 3) the answer to a question whether a *line* complies with '$whatever'
> is --- I believe --- undefined. (It definitely should match '$';
> however comparing with '$whatever' gets us outside the definitions).

supposing that $whatever does parse, it would match
a line with 'whatever' following the end of the line.
by definition there is no such line.

- erik

Re: [9fans] grep (or perhaps regex(6)) question

[Russ Cox]

On Tue, Jul 5, 2011 at 3:56 AM, dexen deVries <dexen.devries@gmail.com> wrote:
> on plan9port:
> % echo foobar | 9 grep '$JUNK'
>  -> foobar
>
> that is, p9p grep matches any line with regular expression in form of
> $ANY_REGEX
>
> why is it so?

It's an ancient bug in the Plan 9 grep, a complication
of the way Ken implemented ^ and $: as aliases for \n (!).
It looks like someone added a workaround for this one
case in the Plan 9 sources since I forked them for
plan9port seven years ago.  I pulled it in.

For fun, try seeing what matches '^$^$' across a
variety of grep implementations.

Russ