On Thu, 19 Jul 2012, Frank Shearar wrote:

> On 18 July 2012 23:23, Levente Uzonyi <leves at elte.hu> wrote:
>> On Wed, 18 Jul 2012, Chris Muller wrote:
>>
>>> Perhaps any package that cares about that preference could set it in
>>> its own preamble.
>>
>>
>> The plan was that we'll turn that preference on by default sometime soon.
>> But there's some cleanup to be done first.
>
> I'd be happy to help bring that day closer by doing some cleanup. Do
> we have a list anywhere?

I don't have a complete list, but here's what I had in my mind:

- remove the per class support, because it's broken and since noone 
reported any issues, it must be unused
- fix Character >> #tokenish to accept $_ if the global preference is 
enabled, something like this should work:

 	^self == $_
 		ifTrue: [ Scanner allowUnderscoreSelectors ]
 		ifFalse: [ self == $: or: [ self isLetter or: [ self isDigit ] ] ]
- fix Symbol >> #precedence to support underscores if the global 
preference is enabled
- clean up/fix String >> #numArgs and String class >> #initialize. It's 
pretty likely that the class variable Tokenish should be reinitialized 
when the preference changes (extract the initialization of the variable to 
a separate method and send it when the preference is changed)


Levente

>
> frank
>
>> Levente
>>
>>
>>>
>>>
>>> On Tue, Jul 17, 2012 at 11:00 AM, Shaun Johnson
>>> <shaun at slugfest.demon.co.uk> wrote:
>>>>
>>>> On 16/07/2012 12:38, commits at source.squeak.org wrote:
>>>>>
>>>>>
>>>>> John McKeon uploaded a new version of FFI-Win32 to project The Inbox:
>>>>> http://source.squeak.org/inbox/FFI-Win32-JohnMcKeon.11.mcz
>>>>>
>>>>> ==================== Summary ====================
>>>>>
>>>>> Name: FFI-Win32-JohnMcKeon.11
>>>>> Author: JohnMcKeon
>>>>> Time: 16 July 2012, 3:33:42.926 pm
>>>>> UUID: 3dda278c-8af6-4c40-a4e9-9cae2cfa574f
>>>>> Ancestors: FFI-Win32-tbn.10
>>>>>
>>>>> add preamble to allow underscores in selectors
>>>>>
>>>>> ==================== Snapshot ====================
>>>>>
>>>>> (PackageInfo named: 'FFI-Win32') preamble: '"Constants have underscores
>>>>> in
>>>>> selectors; set the preference to allow them"
>>>>> Scanner prefAllowUnderscoreSelectors: true'!
>>>>
>>>>
>>>>
>>>> This leave the preference set to true, couldn't this be
>>>> viewed as slightly anti-social?
>>>>
>>>> I suppose this could be reset in the postscript but I don't
>>>> see how to, cleanly, pass the initial state to the postscript.
>>>>
>>>> Shaun.
>>>>
>>>
>>>
>>
>
>