bug in zsh 3.0-pre6?

bug in zsh 3.0-pre6?

[Tomas Gradin]

Hello,

it seems that it has a problem with 8-bit characters immediately following a 
redirection (like |, > etc.).

If I do:

cat>é

..I get:

zsh: Oops. You discovered a bug in hungetc

If I precede the 'é' with a space, everything works.

/tg

Re: bug in zsh 3.0-pre6?

[Zoltan Hidvegi]

> Hello,
> 
> it seems that it has a problem with 8-bit characters immediately following a 
> redirection (like |, > etc.).
> 
> If I do:
> 
> cat>é
> 
> ..I get:
> 
> zsh: Oops. You discovered a bug in hungetc
> 
> If I precede the 'é' with a space, everything works.

Fortunately, this is not a real bug, it's just a missing cast in a debug
test.

Zoltan


*** Src/hist.c	1996/08/12 01:36:46	2.24
--- Src/hist.c	1996/08/13 18:32:08
***************
*** 521,527 ****
  	if (hptr == chline)
  	    zerr("hungetc attempted at buffer start", NULL, 0);
  	else {
! 	    if (hptr[-1] != c)
  		/* Someone probably tried to unget something which is *
  		 * not returned by hgetc.                             */
  		zerr("Oops. You discovered a bug in hungetc", NULL, 0);
--- 521,527 ----
  	if (hptr == chline)
  	    zerr("hungetc attempted at buffer start", NULL, 0);
  	else {
! 	    if (hptr[-1] != (char) c)
  		/* Someone probably tried to unget something which is *
  		 * not returned by hgetc.                             */
  		zerr("Oops. You discovered a bug in hungetc", NULL, 0);

Re: bug in zsh 3.0-pre6?

[Zefram]

>  	if (hptr =3D=3D chline)

Is the broken-printable lossage *really* necessary?  It's a real
nuisance when trying to apply patches.

-zefram

Re: bug in zsh 3.0-pre6?

[Bart Schaefer]

On Aug 13,  8:11pm, Zefram wrote:
} Subject: Re: bug in zsh 3.0-pre6?
}
} >  	if (hptr =3D=3D chline)
} 
} Is the broken-printable lossage *really* necessary?  It's a real
} nuisance when trying to apply patches.

I didn't get any lossage ... the problem appears to be that Zoltan
is sending out

Content-Transfer-Encoding: 8bit

which isn't necessary (his message is all 7bit) but which causes "smart"
ESMTP agents to convert to quoted-printable when gatewaying to an old
SMTP (not 8-bit-clean) agent.

Zoltan didn't do the encoding; it happened along the way somewhere.
But Zoltan could have prevented it by using 7bit as his C-T-E, if he
has that level of control over his mail client.

-- 
Bart Schaefer                             Brass Lantern Enterprises
http://www.well.com/user/barts            http://www.nbn.com/people/lantern

New male in /home/schaefer:
>N  2 Justin William Schaefer  Sat May 11 03:43  53/4040  "Happy Birthday"

Re: bug in zsh 3.0-pre6?

[Richard Coleman]

> I didn't get any lossage ... the problem appears to be that Zoltan
> is sending out
> 
> Content-Transfer-Encoding: 8bit
> 
> which isn't necessary (his message is all 7bit) but which causes "smart"
> ESMTP agents to convert to quoted-printable when gatewaying to an old
> SMTP (not 8-bit-clean) agent.
> 
> Zoltan didn't do the encoding; it happened along the way somewhere.
> But Zoltan could have prevented it by using 7bit as his C-T-E, if he
> has that level of control over his mail client.

The sendmail on our mail machine (where the list is running) is
version 8.7.3.  I imagine that is where the QP converions is
happening.

rc

Re: bug in zsh 3.0-pre6?

[Zoltan Hidvegi]

> >  	if (hptr =3D=3D chline)
> 
> Is the broken-printable lossage *really* necessary?  It's a real
> nuisance when trying to apply patches.

No, quoted-printable is not necessary.  The original mail used plain 8bit
encoding.  The mail server runs sendmail-8.7.3 which asks the destination
machine wether it accepts 8bit mails and if it doesn't sendmail converts
the 8bit message to quoted printable.  But even quoted-printable should be
handled by all MIME aware mail readers.  So it seems that both the MTA and
your mail reader needs some upgrade on your system.

Zoltan

Re: bug in zsh 3.0-pre6?

[Zoltan Hidvegi]

> I didn't get any lossage ... the problem appears to be that Zoltan
> is sending out
> 
> Content-Transfer-Encoding: 8bit
> 
> which isn't necessary (his message is all 7bit) but which causes "smart"
> ESMTP agents to convert to quoted-printable when gatewaying to an old
> SMTP (not 8-bit-clean) agent.
> 
> Zoltan didn't do the encoding; it happened along the way somewhere.
> But Zoltan could have prevented it by using 7bit as his C-T-E, if he
> has that level of control over his mail client.

There was a letter `é' in the message I quoted.  My mail client
(elm-2.4ME+) scans the message for characters with 8th bit set and if it
finds one it automatically uses 8bit, quoted-printable or base64 encoding
(it uses quoted-printable only if I tell it during compile time that my MTA
does not support 8bit, and it uses base64 if more the percentage of 8th bit
set characters is high).

When I read a message, quoted-printable or base64 is seemlessly converted
on the fly by elm.

Zoltan

Re: bug in zsh 3.0-pre6?

[Zefram]

>When I read a message, quoted-printable or base64 is seemlessly converted
>on the fly by elm.

That's odd, with the Elm I use the seam is so big I was compelled to
disable the conversion.  (It invokes metamail, which, not to put too
fine a point on it, sucks rocks.)  But regardless of the method of
display, the message as stored in the mail folder is mangled, and
that's what matters when trying to apply the patch.

But anyway, it's clear that Zoltan is not to blame for this particular
instance of MIME-manglement, and I wish I'd never started this thread.
(I shall simply carry on being annoyed by the occasional message until
I get around to writing a filter to interpret quoted-printable.
metamail doesn't have the capability -- it drops the headers at least,
and can do worse, and I hate to think what it would do with invalid
quoting.)

-zefram

Re: bug in zsh 3.0-pre6?

[Zoltan Hidvegi]

> >When I read a message, quoted-printable or base64 is seemlessly converted
> >on the fly by elm.
> 
> That's odd, with the Elm I use the seam is so big I was compelled to
> disable the conversion.  (It invokes metamail, which, not to put too
> fine a point on it, sucks rocks.)  But regardless of the method of
> display, the message as stored in the mail folder is mangled, and
> that's what matters when trying to apply the patch.

elm-2.4ME+ processes text-only mime messages without calling metamail.
There are also several security fixes in that version and built-in PGP
support.  It is available from ftp://dionysos.fmi.fi/KEH/ but there is a
mirror in ftp://ftp.cs.elte.hu/pub/unix/elm-2.4ME+/ :-).  I apply patches
using |$patch when I read a mail with less.

Zoltan