On Sat, Aug 13, 2011 at 8:41 AM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
2011/8/13 St�phane Ducasse <stephane.ducasse@inria.fr>:
>
> On Aug 12, 2011, at 3:58 PM, Lukas Renggli wrote:
>
>> On 12 August 2011 15:42, St�phane Ducasse <stephane.ducasse@inria.fr> wrote:
>>> I mean that I prefer
>>>
>>>>> Smalltalk tools
>>>>> � � � � � � � �register: ;lkl;k;lk;l
>>>>> � � � � � � � �in: lkl;k;lkl;kl;
>>>
>>> over
>>>
>>>>> Smalltalk tools register: ;lkl;k;lk;l
>>>>> � � � � � � � �in: lkl;k;lkl;kl;
>>>
>>>
>>> because I do not see fast that the message is register: in:
>>
>> Hmm, I still don't understand your example because the $; has semantic
>> meaning and because the �example doesn't even compile :-)
>
>>>>> Smalltalk tools
>>>>> � � � � � � � �register: 'string'
>>>>> � � � � � � � �in: 'hkjjkhkhjh'
>
> :)
>
>
>
>>
>>> and I guess that you are formatting the way I like but the rb formatter may be wrong because
>>
>> You can configure after how how many keyword parts (that is probably
>> too low right now) it should print on one line. Also you can tell it
>> how many characters per line you maximally want (used to be 80, I
>> think i changed the default to 100; but probably should be inifite).
>>
>
> I will play with it.
> I'm really thinking that having a systematic formatting of code within the core would be good.
>

The thing that upsets me most is auto-formatting of literals.

1) the formatter SHALL try to preserve the indenting of array literals
� A 256 elements table organised in a 16x16 fashion can be legible.
� A rough 256 element suit is not.
2) the formatter SHALL preserve the base used to specify a number literal
� If i write 2r1011 it's clearly not my intention to read 11.
� It's the same for number composing a literal array/byte array

In other words, the literals shall better be left unchanged.

+2 raisedTo: SmallInteger maxVal

The second thing that is annoying is that it can become hard to
compare code and track evolutions.
But this one is just a question of having the right tool to perform
the comparison...
I always thought that MC, change list and friends should propose an
option that simply compares scanned tokens and optionnally ignore
comments.

Nicolas

>
>> Lukas
>>
>>>
>>> On Aug 12, 2011, at 1:49 PM, Lukas Renggli wrote:
>>>
>>>> Hmm, I don't get your example.
>>>>
>>>> What I changed is that the formatter does not enforce the useless
>>>> parenthesis around the receiver of a cascade. Can be disabled though.
>>>>
>>>> Also note that there are settings to automatically format code on
>>>> display and/or accept. I use the "on display" mode often to read ugly
>>>> code :-)
>>>>
>>>> Lukas
>>>>
>>>> On 12 August 2011 13:29, St�phane Ducasse <stephane.ducasse@inria.fr> wrote:
>>>>> Yes
>>>>>
>>>>> � � � �Smalltalk tools
>>>>> � � � � � � � �register: ;lkl;k;lk;l
>>>>> � � � � � � � �in: lkl;k;lkl;kl;
>>>>>
>>>>>
>>>>> stef
>>>>> On Aug 12, 2011, at 1:20 PM, Lukas Renggli wrote:
>>>>>
>>>>>>> I always think of this a the Seaside style, but I am not sure if that is a correct attribution.
>>>>>>
>>>>>> Haha, thank you.
>>>>>>
>>>>>> Actually the refactoring browser formatter with the default settings
>>>>>> is quite close to the Seaside style, I just recently submitted some
>>>>>> fixes for cascades. The only thing that I don't like about the
>>>>>> automatic formatter is how the brackets in multiline blocks are
>>>>>> aligned. Anyway, the formatter would style #inject:into: as you
>>>>>> suggest.
>>>>>>
>>>>>> Lukas
>>>>>>
>>>>>> --
>>>>>> Lukas Renggli
>>>>>> www.lukas-renggli.ch
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Lukas Renggli
>>>> www.lukas-renggli.ch
>>>>
>>>
>>>
>>>
>>
>>
>>
>> --
>> Lukas Renggli
>> www.lukas-renggli.ch
>>
>
>
>




--
best,
Eliot