I noticed object.d alias string as invariant(char)[]

But it's a bad choice of the name. string causes a lot name collision in my app.
And as a type. with first char capitalized is a much better choice!!

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/

On 12/14/07, davidl <davidl@126.com> wrote:
>
> I noticed object.d alias string as invariant(char)[]
>
> But it's a bad choice of the name. string causes a lot name collision in my app.

So ... change your app?

Seriously, there's no way Walter could possibly have known what name choices you had made in your app, unless you're suggesting that Walter should be more clairvoyant.

> And as a type. with first char capitalized is a much better choice!!

And do you suppose that this would cause no name clashes at all with anyone's apps? I imagine it would cause /more/. Anyone who's ever made a string class has probably called it "String". To me, the lower case s not only ensures fewer name clashes, but also makes it "feel" like a primitive type (which it sort of almost is - I mean it's just an array, not actually a class or a struct).

Janice Caron Wrote:

> And do you suppose that this would cause no name clashes at all with anyone's apps? I imagine it would cause /more/. Anyone who's ever made a string class has probably called it "String". To me, the lower case s not only ensures fewer name clashes, but also makes it "feel" like a primitive type (which it sort of almost is - I mean it's just an array, not actually a class or a struct).

Agreed :)

On an unrelated note, module names starting with uppercase letters (like those in Tango) have always turned me off -- feels too Java-ish to me :P (I remember reading somewhere in the forum that this is done for good reasons, though)

Colin Huang wrote:
> Janice Caron Wrote:
> 
>> And do you suppose that this would cause no name clashes at all with
>> anyone's apps? I imagine it would cause /more/. Anyone who's ever made
>> a string class has probably called it "String". To me, the lower case
>> s not only ensures fewer name clashes, but also makes it "feel" like a
>> primitive type (which it sort of almost is - I mean it's just an
>> array, not actually a class or a struct).
> 
> Agreed :)
> 
> On an unrelated note, module names starting with uppercase letters (like those in Tango) have always turned me off -- feels too Java-ish to me :P (I remember reading somewhere in the forum that this is done for good reasons, though)

Tango names _Packages_ with with capitalized words, but not the modules.  This allows you to have a module that's the same name as a package. Like "String.string".  "String" is a package/directory and "string.d" is a file in that directory.

--bb

Bill Baxter Wrote:

> Colin Huang wrote:
> > Janice Caron Wrote:
> > 
> >> And do you suppose that this would cause no name clashes at all with anyone's apps? I imagine it would cause /more/. Anyone who's ever made a string class has probably called it "String". To me, the lower case s not only ensures fewer name clashes, but also makes it "feel" like a primitive type (which it sort of almost is - I mean it's just an array, not actually a class or a struct).
> > 
> > Agreed :)
> > 
> > On an unrelated note, module names starting with uppercase letters (like those in Tango) have always turned me off -- feels too Java-ish to me :P (I remember reading somewhere in the forum that this is done for good reasons, though)
> 
> Tango names _Packages_ with with capitalized words, but not the modules.
>   This allows you to have a module that's the same name as a package.
> Like "String.string".  "String" is a package/directory and "string.d" is
> a file in that directory.
> 
> --bb

Perhaps I'm misunderstanding here, but have a look at the following code from tango-bin-win32-CURRENT-dmd.1.022 (in file \import\tango\text\String.d):

module tango.text.String;

Isn't this a module with a capitalized name? I'm confused ...

On Fri, 14 Dec 2007 07:18:43 -0000, Janice Caron <caron800@googlemail.com> wrote:

> On 12/14/07, davidl <davidl@126.com> wrote:
>>
>> I noticed object.d alias string as invariant(char)[]
>>
>> But it's a bad choice of the name. string causes a lot name collision in
>> my app.
>
> So ... change your app?
>
> Seriously, there's no way Walter could possibly have known what name
> choices you had made in your app, unless you're suggesting that Walter
> should be more clairvoyant.
>
>> And as a type. with first char capitalized is a much better choice!!
>
> And do you suppose that this would cause no name clashes at all with
> anyone's apps? I imagine it would cause /more/. Anyone who's ever made
> a string class has probably called it "String". To me, the lower case
> s not only ensures fewer name clashes, but also makes it "feel" like a
> primitive type (which it sort of almost is - I mean it's just an
> array, not actually a class or a struct).

Actually if D had namespaces he wouldn't need to be clairvoyant. Everything
would be safely tucked away in namespace std:: and only imported if the programmer
asked. IIRC the argument against them was simplicity but them to me to be an
essential grouping structure at the architecture level. Of course the need for them
can be delayed a fair while in the land of modules.

On Fri, 14 Dec 2007 10:14:40 +0100, Colin Huang <colin.hwong@gmail.com> wrote:

> Bill Baxter Wrote:
>
>> Colin Huang wrote:
>> > Janice Caron Wrote:
>> >
>> >> And do you suppose that this would cause no name clashes at all with
>> >> anyone's apps? I imagine it would cause /more/. Anyone who's ever  
>> made
>> >> a string class has probably called it "String". To me, the lower case
>> >> s not only ensures fewer name clashes, but also makes it "feel" like  
>> a
>> >> primitive type (which it sort of almost is - I mean it's just an
>> >> array, not actually a class or a struct).
>> >
>> > Agreed :)
>> >
>> > On an unrelated note, module names starting with uppercase letters  
>> (like those in Tango) have always turned me off -- feels too Java-ish to me :P (I remember reading somewhere in the forum that this is done for good reasons, though)
>>
>> Tango names _Packages_ with with capitalized words, but not the modules.
>>   This allows you to have a module that's the same name as a package.
>> Like "String.string".  "String" is a package/directory and "string.d" is
>> a file in that directory.
>>
>> --bb
>
> Perhaps I'm misunderstanding here, but have a look at the following code
> from tango-bin-win32-CURRENT-dmd.1.022 (in file \import\tango\text\String.d):
>
> module tango.text.String;
>
> Isn't this a module with a capitalized name? I'm confused ...

Seems Bill was just mixing things up. From a quick look at my Tango installation, it seems packages are lowercase, modules are capitalized.

Colin Huang wrote:
> Bill Baxter Wrote:
> 
>> Colin Huang wrote:
>>> Janice Caron Wrote:
>>>
>>>> And do you suppose that this would cause no name clashes at all with
>>>> anyone's apps? I imagine it would cause /more/. Anyone who's ever made
>>>> a string class has probably called it "String". To me, the lower case
>>>> s not only ensures fewer name clashes, but also makes it "feel" like a
>>>> primitive type (which it sort of almost is - I mean it's just an
>>>> array, not actually a class or a struct).
>>> Agreed :)
>>>
>>> On an unrelated note, module names starting with uppercase letters (like those in Tango) have always turned me off -- feels too Java-ish to me :P (I remember reading somewhere in the forum that this is done for good reasons, though)
>> Tango names _Packages_ with with capitalized words, but not the modules.   This allows you to have a module that's the same name as a package. Like "String.string".  "String" is a package/directory and "string.d" is a file in that directory.
>>
>> --bb
> 
> Perhaps I'm misunderstanding here, but have a look at the following code
> from tango-bin-win32-CURRENT-dmd.1.022 (in file \import\tango\text\String.d):
> 
> module tango.text.String;
> 
> Isn't this a module with a capitalized name? I'm confused ...

Doh!  I got it backwards.  Sorry.  Anyway the modules and packages have different capitalization so that a module and package can have the same name.

--bb

--bb

On Fri, 14 Dec 2007 15:09:22 +0800, davidl wrote:

> I noticed object.d alias string as invariant(char)[]
> 
> But it's a bad choice of the name. string causes a lot name collision in
> my app.
> And as a type. with first char capitalized is a much better choice!!

This suggestion is a little late I guess, but I try to never have my own identifiers be either valid English words or be in a single case, just so I can avoid things like 'string' popping up in D.

-- 
Derek Parnell
Melbourne, Australia
skype: derek.j.parnell

davidl wrote:
> 
> I noticed object.d alias string as invariant(char)[]
> 
> But it's a bad choice of the name. string causes a lot name collision in my app.
> And as a type. with first char capitalized is a much better choice!!
> 

Primitives get lower-case identifiers. That's all there is to it.