Well, Squeak still accept #: but it already prints #':' because we
refactored Scanner class>>isLiteralSymbol: some years ago
We also changed String>>numArgs so that #':if:then:else:' numArgs = -1
It would be logical to finish the clean-up.
Only the possibility to load historical code should retain us if it really
matters
(but it's hard to tell without analyzing large code bases).

That's one thing we should always remember: adding a small extension to the
syntax is very easy to do.
Removing it is much much harder (for social reasons).

Pharo 3.0 does not accept #: #:x anymore since OpalCompiler switch.


2013/9/12 Bert Freudenberg <bert at freudenbergs.de>

>
> On 2013-09-12, at 23:03, Levente Uzonyi <leves at elte.hu> wrote:
>
> > On Thu, 12 Sep 2013, Bert Freudenberg wrote:
> >
> >> On 2013-09-11, at 23:14, Nicolas Cellier <
> nicolas.cellier.aka.nice at gmail.com> wrote:
> >>
> >>>      After verification, xColon is not so rare, it is invoked for
> every block argument...
> >>>      But we can safely replace (token := self step asSymbol) with
> (token := String with: self step), no problem.
> >>> We can even change it to (token := self step) if we are willing to
> remove last scories of alternate selectors beginning with $:
> >>> Indeed, #: and #:x are obsolete squeakism which would not compile in
> other dialects (#':' and #':x' is the universally correct syntax).
> >> Well, all Squeak-derived dialects still do support it. Pharo changed
> the printString of #: to #':', while Squeak and Cuis still print #:. And
> they all accept #:, #:x, and #:x:::.
> >
> > You mean you can still use #: or #:x as a selector in Squeak, Cuis or
> Pharo? I'm pretty sure that was removed long ago from Squeak, and I doubt
> the other dialects support them.
>
>
> No, I mean they recognize the symbol syntax.
>
> - Bert -
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20130913/a4c26f28/attachment.htm