zsh dies (3.1.5-pws-16)

zsh dies (3.1.5-pws-16)

[Tatsuo Furukawa]

Hello, zsh developers.  I found another bug.


1. Start new zsh

    $ zsh

2. Check that COLUMNS environment variable is defined.

    $ env | grep COLUMNS   
    COLUMNS=80

3. Remove COLUMNS

    $ unset COLUMNS

4. Execute "env" command.

    $ env

5. zsh dies. (ToT)

    zsh: segmentation fault  zsh


I tested the following situation:

Dies:
    zsh:    3.1.5-pws-16
    OS:     Linux/Slackware 3.1

OK:
    zsh:    3.0.6-pre-2
    OS:     Linux/Slackware 3.1


I didn't make patch to fix this problem.  Because, Bart Schaefer said
that zsh 3.1.X's resize code is older than 3.0.X's (NOTE 1).  And,
3.0.6-pre-2 does not have the problem.  So, I think 3.1.X's resize
code should be changed into 3.0.X's.  (I can help to making patch!)


NOTE1:
>> From: "Bart Schaefer" <schaefer@brasslantern.com>
>> Date: Sat, 24 Oct 1998 11:43:34 -0700
>> Message-Id: <981024114334.ZM30186@candle.brasslantern.com>
>> Subject: Re: [3.0.5] patch for resize problem
>> X-Mailing-List: <zsh-workers@math.gatech.edu> archive/latest/4447

-- 
Tatsuo Furukawa  (frkwtto@osk3.3web.ne.jp)

Re: zsh dies (3.1.5-pws-16)

[Peter Stephenson]

Tatsuo Furukawa wrote:
> Hello, zsh developers.  I found another bug.
> 
> I didn't make patch to fix this problem.  Because, Bart Schaefer said
> that zsh 3.1.X's resize code is older than 3.0.X's (NOTE 1).  And,
> 3.0.6-pre-2 does not have the problem.  So, I think 3.1.X's resize
> code should be changed into 3.0.X's.  (I can help to making patch!)

This is so complicated I'm unhappy about ripping the code from one to the
other without being told what's really going on, if anyone knows.

- What happened to the patch 4447 (from Bart) that was supposed to make
3.0.5 believe SIGWINCH signals?  Shouldn't that be in 3.0.6?

- Where did the code change between 3.0 and 3.1 originate anyway (and why)?
Searching for adjustwinsize in the zsh-workers archive doesn't produce a
patch to explain it, so it's hard to know what's a step forward and what's
a step backward.  It's all mixed in with little changes like moving the
definitions of the LINES and COLUMNS variables from hashtable.h to
params.c.

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy

Re: zsh dies (3.1.5-pws-16)

[Bart Schaefer]

On Apr 29, 10:17am, Peter Stephenson wrote:
} Subject: Re: zsh dies (3.1.5-pws-16)
}
} Tatsuo Furukawa wrote:
} > Hello, zsh developers.  I found another bug.
} > 
} > I didn't make patch to fix this problem.  Because, Bart Schaefer said
} > that zsh 3.1.X's resize code is older than 3.0.X's (NOTE 1).  And,
} > 3.0.6-pre-2 does not have the problem.  So, I think 3.1.X's resize
} > code should be changed into 3.0.X's.  (I can help to making patch!)
} 
} This is so complicated I'm unhappy about ripping the code from one to the
} other without being told what's really going on, if anyone knows.
} 
} - What happened to the patch 4447 (from Bart) that was supposed to make
} 3.0.5 believe SIGWINCH signals?  Shouldn't that be in 3.0.6?

What happened to it was that I never applied it to my local copy of zsh
because I wasn't having the problem it fixed, and I was at that time
leaving it up to Zefram to decide whether to include patches in the
official versions.  Consequently, I missed it when making up the -pre-X
patches.

I'll get it into -pre-3, as it looks like there needs to be one more
round of testing anyway.

} - Where did the code change between 3.0 and 3.1 originate anyway (and why)?

As best I can tell, the change was put into 3.0 after the split for 3.1
and then simply never propagated into 3.1.  Cf. zsh-workers 4919, 5301.

} It's all mixed in with little changes like moving the
} definitions of the LINES and COLUMNS variables from hashtable.h to
} params.c.

I suspect that's why Zoltan and/or Zefram never got around to duplicating
this and other changes.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com