Re: Shell variable tables:setting value for elemet 1 overwrites element 0

Re: Shell variable tables:setting value for elemet 1 overwrites element 0

[Peter Stephenson]

wojciech.szenajch@bull.com.pl wrote:
> If this is the case, then any operation performed on element 0 should
> return error message. This is misleading that operations like
> TB[0]=zero; echo ${TB[0]} are accepted.

Yes, it was probably the wrong decision to accept [0] in this case.
I hope no one's relying on it, but you can't be sure.

> I wonder why this zsh behaviour is kept, if all other popular shells number
> arrays from 0.

The array feature was originally based around csh rather than ksh.  Accepting
0 as an alias for 1 dates from this time, I think.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

**********************************************************************

Re: Shell variable tables:setting value for elemet 1 overwrites element 0

[Bart Schaefer]

On May 25,  1:05pm, Peter Stephenson wrote:
} Subject: Re: Shell variable tables:setting value for elemet 1 overwrites e
}
} wojciech.szenajch@bull.com.pl wrote:
} > If this is the case, then any operation performed on element 0 should
} > return error message. This is misleading that operations like
} > TB[0]=zero; echo ${TB[0]} are accepted.
} 
} Yes, it was probably the wrong decision to accept [0] in this case.
} I hope no one's relying on it, but you can't be sure.

I don't think there's any reason to allow *assignment* to index zero,
except perhaps as a way to accomplish array=(newelement $array).  The
dereference of $array[0] is a different matter.

Re: Shell variable tables:setting value for elemet 1 overwrites element 0

[Peter Stephenson]

Bart Schaefer wrote:
> On May 25,  1:05pm, Peter Stephenson wrote:
> } Subject: Re: Shell variable tables:setting value for elemet 1 overwrites e
> }
> } wojciech.szenajch@bull.com.pl wrote:
> } > If this is the case, then any operation performed on element 0 should
> } > return error message. This is misleading that operations like
> } > TB[0]=zero; echo ${TB[0]} are accepted.
> } 
> } Yes, it was probably the wrong decision to accept [0] in this case.
> } I hope no one's relying on it, but you can't be sure.
> 
> I don't think there's any reason to allow *assignment* to index zero,
> except perhaps as a way to accomplish array=(newelement $array).  The
> dereference of $array[0] is a different matter.

Unfortunately it doesn't seem to be that easy to change one but not the
other; retrieving the index is fairly well buried.  Extra flags could do
this but the proliferation of getvalue() and fetchvalue() in the code
makes it rather messy to test for failure in all the right places.
Also, getindex(), which would have to perform the test, is called
multiple times deep inside the code.

Generally, the whole structure of the code is wrong to be able to solve
this sort of problem simply.  The current logic is to try something
involving a deep hierarchy of code, and if it fails, try something else.
Handling error cases systematically in code like this is very difficult.

There is actually a good reason for the structure, however, namely a
horrendous degree of context sensitivity in subscripts... if $foo exists
and is an array or scalar, foo[0]=bar should fail; if foo exists but is a
hash it should succeed; if foo doesn't exist, it needs to be created
as an array but this should fail when the assignment is made.  And so on.

I don't think this comes as a great surprise.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

**********************************************************************

Shell variable tables:setting value for elemet 1 overwrites element 0

[wojciech.szenajch]

Hi,

I did the following test for AIX 4.3.3/5.2 ksh, Linux: bash, and pdksh:

$ TB[0]=zero
$ TB[1]=one
$ echo ${TB[0]}
zero
$ echo ${TB[1]}
one


For zsh 4.2.0  (SuSE 9.1, Red Hat ES 4.0)
I did the following:

$ zsh --version
zsh 4.2.0 (i386-suse-linux)    or zsh 4.2.0 (i386-redhat-linux-gnu)
$ zsh
$ TB[0]=zero
$ echo ${TB[0]}
zero     <-------------------------------- correct result
$ TB[1]=one
$ echo ${TB[0]}
one     <-------------------------------- incorrect result
$ echo ${TB[1]}
one


Best Regards
Wojciech Szenajch

Re: Shell variable tables:setting value for elemet 1 overwrites element 0

[Peter Stephenson]

wojciech.szenajch@bull.com.pl wrote:
> $ zsh --version
> zsh 4.2.0 (i386-suse-linux)    or zsh 4.2.0 (i386-redhat-linux-gnu)
> $ zsh
> $ TB[0]=zero
> $ echo ${TB[0]}
> zero     <-------------------------------- correct result
> $ TB[1]=one
> $ echo ${TB[0]}
> one     <-------------------------------- incorrect result
> $ echo ${TB[1]}
> one

This is correct.  "setopt ksh_arrays" or start up under the alias sh
or ksh to get the other behaviour.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

**********************************************************************