On 13.03.2015, at 03:34, Levente Uzonyi <leves at elte.hu> wrote:

> On Fri, 13 Mar 2015, Tobias Pape wrote:
> 
>> Hey,
>> 
>> 
>> On 13.03.2015, at 00:34, commits at source.squeak.org wrote:
>> 
>>> Tobias Pape uploaded a new version of Compiler to project The Trunk:
>>> http://source.squeak.org/trunk/Compiler-topa.295.mcz
>>> 
>>> ==================== Summary ====================
>>> 
>>> Name: Compiler-topa.295
>>> Author: topa
>>> Time: 13 March 2015, 1:34:26.208 am
>>> UUID: f2727a06-5024-4534-bf38-934bcdad9bf9
>>> Ancestors: Compiler-bf.294
>>> 
>>> Fix underscore message names when the #parseSelector: api of Parser is used and, hence, no encoder is available.
>>> 
>>> =============== Diff against Compiler-bf.294 ===============
>>> 
>>> Item was changed:
>>> ----- Method: Parser>>allowUnderscoreAssignments (in category 'private') -----
>>> allowUnderscoreAssignments
>>> 	"Query class + preference"
>>> + 	^ (encoder ifNotNil: [:e | e classEncoding allowUnderscoreAssignments])
>>> + 		ifNil: [super allowUnderscoreAssignments]!
>>> - 	^encoder classEncoding allowUnderscoreAssignments
>>> - 		ifNil:[super allowUnderscoreAssignments]!
>> 
>> To be frank, this is a bit of a hack, because the class's opinion on #allowUnderscoreAssignments
>> is regrettably ignored. But this is better than an MNU on nil I think…
> 
> That option - letting a class decide if its selectors may or may not contain underscores - should be nuked. It doesn't work, and it won't work.
> 

Why the former and why the latter?

Best
	-Tobias

> Levente
> 
>> 
>> Best
>> 	-Tobias