Re: [PATCH 1/3] Extended ksh compatibility: namespace parameter syntax

[Mikael Magnusson <mikachu@xxxxxxxxx>]

On 5/20/23, Phil Pennock <zsh-workers+phil.pennock@xxxxxxxxxxxx> wrote:
> On 2023-05-19 at 23:54 -0700, Bart Schaefer wrote:
>> On Fri, May 19, 2023 at 7:06 PM Phil Pennock
>> <zsh-workers+phil.pennock@xxxxxxxxxxxx> wrote:
>> > AFAICT, in zsh, .a.b=c is assigning to var b in namespace a, while
>> > a.b=c
>> > is setting the b attribute of the compound variable a.
>>
>> As presently implemented, there's no difference.  The latter is
>> supported by the parser but should be considered a reserved syntax;
>> some sort of compound variable feature is a possible future
>> application.
>
> Sorry, I thought I'd fixed that typo before sending.
>
> In _ksh_.
>
> I played with ksh to evaluate the behavior, which didn't match what was
> said upthread, which is why I spoke up to clarify the distinction.
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~8< ksh >8~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> % ksh
> $ echo ${.sh.version}
> Version AJM 93u+ 2012-08-01
> $ PS1='ksh$ '
> ksh$ foo=( alpha=first beta=second ) # compound var
> ksh$ namespace bar { alpha=one beta=two ; }
> ksh$ echo ${foo}
> ( alpha=first beta=second )
> ksh$ echo ${.foo}
>
> ksh$ echo ${bar}
>
> ksh$ echo ${.bar}
> alpha beta
> ksh$ foo.gamma=third
> ksh$ echo $foo
> ( alpha=first beta=second gamma=third )
> ksh$ .bar.gamma=three
> ksh$ echo ${.bar}
> alpha beta
> ksh$ namespace bar { echo $gamma; }
> three
> ksh$ echo ${.bar.gamma}
> three
> ksh$ echo ${@foo}
> typeset -C
> ksh$ echo ${@bar}
>
> ksh$ echo ${@.bar}
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~8< ksh >8~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> I don't know what's going on with `echo ${.bar}`, it's undocumented in
> my man-page and the inability to use ${@vname} to get details about it
> suggests perhaps it shouldn't expand?
> (  ksh ${@vname} ~~ zsh ${(t)vname}  )
>
> So, to match the intent of ksh as I understand it:
>  * .identifier.x is always treating identifier as a namespace
>  * identifier.x is always treating identifier as a compound var
>  * this is consistent across assignment LHS and expansion

Does this mean compound vars can't be namespaced?

-- 
Mikael Magnusson