Re: status on $[arith] for eval arith vsl $((arith))??

[Elliott Forney]

My two cents, would be to add a bit to the man page that says something like:

-------
For historical reasons, the following format is also supported for
arithmetic evaluation:

$[expression]

Note, however, that this format is non-standard and that
$((expression)) is preferred.
-------

At least that way there won't be an undocumented feature.

Thanks,
  Elliott Forney

On Mon, Apr 9, 2012 at 8:47 PM, Chet Ramey <chet.ramey@case.edu> wrote:
> On 4/9/12 9:07 PM, Linda Walsh wrote:
>>
>>
>> Maarten Billemont wrote:
>>
>>> Any particular reason for not removing old undocumented functionality,
>>> or is that mostly the nature of this beast - dragging along and maintaining
>>> ancient code for the sake of compatibility?
>
>>
>> So 'yesturday' is "ancient" for you?... that's really means something.
>
> You have a peculiar view of history (spelling and grammar, too, but we'll
> leave that aside).
>
>>
>> In doing a scan over my /usr partition,
>> I see MANY examples in bash 4.1 in it's examples of using $[] -- I would
>> hardly call 4.1 "ancient".
>
> True, I never went back and changed the examples.
>
>>
>>
>> Other packages that make use of the syntax:
>>
>> * wondershaper
>> * cifs file system
>> * alsa (sound)
>> * fonts-config   (this is a new project within the past few years)
>> * QT4
>> * GoogleBrowser (chromium)
>> * RPM
>> * YP
>> * PM Utils
>> * RPMrebuild
>> * iproute2 (almost all modern internet functions on linux)...
>> * dhcp-client
>> - (zsh -- not a real example, but might become alternate system shell if bash
>>  killed $[] support)
>> * Opera - ? (has comment "TODO use $(()) instead of $[] whenever
>> possible;...) "whenever possible??"
>> * samba
>>    and a HUGE number in
>> ** linux-kernel -- all over the place...
>>
>> At that point, I was getting too many to keep up with ... so I stopped
>> searching...
>>
>> $[] has is -- I would bet, Universally, used MORE than $(())...
>
> I believe you'd lose, but it's unprovable either way.  Consider the fact,
> though, that bash and zsh support $[...], ksh93 and dash support only
> $((...)), but that all four support $((...)).
>
>> Chet -- you should get back to the posix folks and tell them posix is to be
>> 'descriptive of usage' (their words), not prescriptive.   Just because ksh
>> did it differently from everyone else's usage doesn't mean they should go
>> with that syntax...
>
> You have got to be kidding.  Don't you realize you're talking about
> decisions that are nearly 20 years old?  That $[...] was a Posix
> invention that only ever appeared in P1003.2d9?  That the $((...))
> syntax was adopted officially for P1003.2-1992?  That's 1992.  Twenty
> years ago.
>
> Bash, and later zsh, implemented $[...] because there was no other
> syntax at the time, and to gain some operational experience with
> arithmetic expansion in the shell.  Bash-1.14 (the oldest I have
> readily available) lists both forms of arithmetic expansion, but by
> the time bash-2.0 was released in 1995, the manual referred only to
> the $((...)) form.  That's 17 years ago.  Hardly "yesterday".
>
> Now, it's hardly any problem to keep dragging the $[...] syntax along.
> It takes only a few dozen bytes of code.  I have no plans to remove it.
> But let's not kid ourselves: it's revisionist history to think that
> $[...] was widespread before Posix so callously stamped it out.
>
>> Geez.
>
> Indeed.
>
> --
> ``The lyf so short, the craft so long to lerne.'' - Chaucer
>                 ``Ars longa, vita brevis'' - Hippocrates
> Chet Ramey, ITS, CWRU    chet@case.edu    http://cnswww.cns.cwru.edu/~chet/
>