Selectors with underscores

> Pharo allows you to use underscores in method selectors.
> This is currently in Pharo 1.1 and not 1.0.
>
> Does that makes sense to have in Squeak trunk/Cuis too?
>
> Try:
>
> (HTTPSocket httpGet:
> 'http://www.assembla.com/spaces/SqueakDBX/documents/dvVhPSvuCr3OqXeJe5aVNr/dow
> nload/underscore.cs')   readStream fileIn
>
> And you can implement a method
>
>    foo_test
>       ^'Works'
>
> Makes sense in conjunction with Glorp (when a database column
> mapped to a class has underscores in it) or FFI (where one
> can name methods like C constants, ...), ...
>
> If it would be in all base images the Glorp loader wouldnt
> have to patch these.
>
> Thx
> Bye

> I have an experimental patch to MC1.5 that converts
> underscore assignments to := at load time, if the compiler
> raises a SyntaxError about them. How it works:
>
> - Parser throws a syntax error
> - MC catches the error
> - MC uses the parser to find underscore assignments (and not
>    underscores in strings/comments/names) and replace them
> - MC retries compiling the method
> - If a syntax error is raised again, MC lets it thru and you see
>    the Syntax Error dialog box
> - If any methods were changed during load, MC tells you about
>    them so you can review them and save the package
>
> Note that it's triggered by a SyntaxError, so this will not be
> done if the preference is to allow underscore assignments.
>
> Also, it only fixes things the parser things are disallowed
> underscore assignments, so it is quite conservative.
>
>
> I use this to prepare packages for loading into Cobalt (where
> underscores assignments are disallowed by default)
>

> 2010/3/11 Torsten Bergmann <[hidden email]>:
>> Pharo allows you to use underscores in method selectors.
>> This is currently in Pharo 1.1 and not 1.0.
>>
>> Does that makes sense to have in Squeak trunk/Cuis too?
>
> No, this is *not* something I'd like to have in Squeak. It's a
> potential disaster in the long run where either we will be forced to
> standardize code in some way or live with a basic image/trunk/etc and
> very weird variable names mixing camel_CameANd_whAtnot beyond the
> usage of Glorp.
>

> On 11 March 2010 16:47, Ian Trudel <[hidden email]> wrote:
>> 2010/3/11 Torsten Bergmann <[hidden email]>:
>>> Pharo allows you to use underscores in method selectors.
>>> This is currently in Pharo 1.1 and not 1.0.
>>>
>>> Does that makes sense to have in Squeak trunk/Cuis too?
>>
>> No, this is *not* something I'd like to have in Squeak. It's a
>> potential disaster in the long run where either we will be forced to
>> standardize code in some way or live with a basic image/trunk/etc and
>> very weird variable names mixing camel_CameANd_whAtnot beyond the
>> usage of Glorp.
>>
>
> I don't see any disaster with it. Take another languages, does a
> presence of underscore in names makes them
> more beatiful/horrible above anything else? I don't think so.
>

>> o_O
>> (how appropriate!)
>>
>> Ian.
>> --
>> http://mecenia.blogspot.com/
>>
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
>

> 2010/3/11 Igor Stasenko <[hidden email]>:
>> On 11 March 2010 16:47, Ian Trudel <[hidden email]> wrote:
>>> 2010/3/11 Torsten Bergmann <[hidden email]>:
>>>> Pharo allows you to use underscores in method selectors.
>>>> This is currently in Pharo 1.1 and not 1.0.
>>>>
>>>> Does that makes sense to have in Squeak trunk/Cuis too?
>>>
>>> No, this is *not* something I'd like to have in Squeak. It's a
>>> potential disaster in the long run where either we will be forced to
>>> standardize code in some way or live with a basic image/trunk/etc and
>>> very weird variable names mixing camel_CameANd_whAtnot beyond the
>>> usage of Glorp.
>>>
>>
>> I don't see any disaster with it. Take another languages, does a
>> presence of underscore in names makes them
>> more beatiful/horrible above anything else? I don't think so.
>>
>
> To my understanding, there is no need of _ in selectors, BUT for
> hooking external procedures.
> It's painful to have to remember two different selectors when you
> heaviily rely on FFI (OpenGL is a good example indeed).
>

> On 11 March 2010 22:35, Nicolas Cellier
> <[hidden email]> wrote:
>> 2010/3/11 Igor Stasenko <[hidden email]>:
>>> On 11 March 2010 16:47, Ian Trudel <[hidden email]> wrote:
>>>> 2010/3/11 Torsten Bergmann <[hidden email]>:
>>>>> Pharo allows you to use underscores in method selectors.
>>>>> This is currently in Pharo 1.1 and not 1.0.
>>>>>
>>>>> Does that makes sense to have in Squeak trunk/Cuis too?
>>>>
>>>> No, this is *not* something I'd like to have in Squeak. It's a
>>>> potential disaster in the long run where either we will be forced to
>>>> standardize code in some way or live with a basic image/trunk/etc and
>>>> very weird variable names mixing camel_CameANd_whAtnot beyond the
>>>> usage of Glorp.
>>>>
>>>
>>> I don't see any disaster with it. Take another languages, does a
>>> presence of underscore in names makes them
>>> more beatiful/horrible above anything else? I don't think so.
>>>
>>
>> To my understanding, there is no need of _ in selectors, BUT for
>> hooking external procedures.
>> It's painful to have to remember two different selectors when you
>> heaviily rely on FFI (OpenGL is a good example indeed).
>>
>
> First, i think we should answer to a simple question: why many other
> computer languages allow underscores in their syntax as a
> valid character for identifiers.
> As to me, the answer is obvious: in natural languages we using space
> to separate two words, and since parsers can't cope (or can, but its
> highly impractical) with identifiers which having spaces:
>
> my variable := foo.
>
> an underscore is a good compromise between natural language and
> computer language to separate two words,
> while treat it as a single identifier:
>
> my_variable := foo.
>
> I came to smalltalk not long ago, and having over 15+ years of
> programming experience.
> I am always used underscores in this way, and never (or occasionally) like this:
>
> _abdomination := foo.
>
>
> Smalltalk spirit is a language which designed to remove unnecessary
> barriers between humand and computer - easy to learn and easy to
> express yourself, without need to be a highly educated
> genius who can translate a natural human language to a cryptic sequece
> of instructions which computer can understand.
> And allowing underscores in identifiers, as to me, is highly coherent
> with that spirit, because it this makes us easier to express
> ourselves.
>
>> Nicolas
>>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
>