www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - private alias ... isn't

reply David Medlock <noone nowhere.com> writes:
In several of my utility libraries I have declared :

private alias char[] string;

I don't even type char[] any longer, I always type string.

Every so often I run into conflicts

alias textures.string conflicts with trees.string ...
(several lines like this)

I can usually monkey around with various imports, but its a PITA.
I don't see any possible benefit for this behavior, if it is intended.

*** If it is a bug, ignore the rest of this post.

Names serve a much larger purpose than locating a variable or method, 
they are *abstractions*.  Why then would a private member be available 
elsewhere?

When I type private I expect that declaration to be removed from the 
namespace when the parser leaves the file.  Thats what *private* 
means(the English word).  Without power to hide some abstractions from 
others we leave the world of 'logic' and enter the 'zeros and ones' 
world again.  Black boxes shouldn't be clear plexiglass.

-puzzling
DavidM
Jun 22 2006
next sibling parent Derek Parnell <derek nomail.afraid.org> writes:
On Thu, 22 Jun 2006 23:45:59 -0400, David Medlock wrote:


 In several of my utility libraries I have declared :
 
 private alias char[] string;
 
 I don't even type char[] any longer, I always type string.
 
 Every so often I run into conflicts
 
 alias textures.string conflicts with trees.string ...
 (several lines like this)
 
 I can usually monkey around with various imports, but its a PITA.
 I don't see any possible benefit for this behavior, if it is intended.
 
 *** If it is a bug, ignore the rest of this post.
 
 Names serve a much larger purpose than locating a variable or method, 
 they are *abstractions*.  Why then would a private member be available 
 elsewhere?
 
 When I type private I expect that declaration to be removed from the 
 namespace when the parser leaves the file.  Thats what *private* 
 means(the English word).  Without power to hide some abstractions from 
 others we leave the world of 'logic' and enter the 'zeros and ones' 
 world again.  Black boxes shouldn't be clear plexiglass.
 
 -puzzling


This could be any other instance where there is a distinction between
"visibility" and "accessibility".

A private alias may be visible from other modules even though it cannot be
used by that other module.


However, after just testing that theory out, it seems that there is no such
beastie as a 'private alias'. The 'private' qualifier is simply ignored.

Pity, because the concept is sound, IMHO.

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
"Down with mediocrity!"
23/06/2006 2:49:54 PM
Jun 22 2006
prev sibling parent reply Lionello Lunesu <lio lunesu.remove.com> writes:
I, too, always create an "alias char[] string"! If only to prevent the 
barely readable: char[][char[]] string2string;

And I think it wouldn't hurt if we include the following aliases in 
Phobos, Ares:

alias char[] string;
alias wchar[] wstring;
alias dchar[] dstring;

L.

(And, perhaps, declare a tchar to be either char or wchar, and tstring a 
tchar[], much like windows' TCHAR and LPTSTR; indeed we'd get two built 
libraries, but I think the .lib on windows should always use wchar and 
the .a on linux should always use char.)
Jun 23 2006
next sibling parent David Medlock <noone nowhere.com> writes:
Lionello Lunesu wrote:

 I, too, always create an "alias char[] string"! If only to prevent the 
 barely readable: char[][char[]] string2string;
 
 And I think it wouldn't hurt if we include the following aliases in 
 Phobos, Ares:
 
 alias char[] string;
 alias wchar[] wstring;
 alias dchar[] dstring;
 
 L.
 
 (And, perhaps, declare a tchar to be either char or wchar, and tstring a 
 tchar[], much like windows' TCHAR and LPTSTR; indeed we'd get two built 
 libraries, but I think the .lib on windows should always use wchar and 
 the .a on linux should always use char.)


YES please.

-DavidM
Jun 23 2006
prev sibling parent reply =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Lionello Lunesu wrote:


 And I think it wouldn't hurt if we include the following aliases in 
 Phobos, Ares:
 
 alias char[] string;
 alias wchar[] wstring;
 alias dchar[] dstring;


I think Walter preferred showing the array nature of strings. Then again 
he also preferred showing the bit nature of bools, things change... :-)

Maybe the time has come for offering a more "readable" alias for it ?

I used "str" myself, partly because it was shorter and partly to avoid 
mixing it up with "std::string" in my head. (since it is not a class)

http://www.prowiki.org/wiki4d/wiki.cgi?CharsAndStrs#SuggestedTypealiases

--anders
Jun 25 2006
parent Lionello Lunesu <lio lunesu.remove.com> writes:
Anders F Björklund wrote:

 Lionello Lunesu wrote:
 

 And I think it wouldn't hurt if we include the following aliases in 
 Phobos, Ares:

 alias char[] string;
 alias wchar[] wstring;
 alias dchar[] dstring;


 
 I think Walter preferred showing the array nature of strings. Then again 
 he also preferred showing the bit nature of bools, things change... :-)


char[] a;
It looks like an array of characters, but a[0] is NOT necessarily the 
first character, so that kind of reasoning can easily backfire.


 Maybe the time has come for offering a more "readable" alias for it ?
 
 I used "str" myself, partly because it was shorter and partly to avoid 
 mixing it up with "std::string" in my head. (since it is not a class)


I prefer short names for temporary variables: "string str;". And calling 
the alias "string" might not be a bad idea: you can mimic std::string's 
interface by creating all of its members global and passing "string" as 
first parameter. They can then be called as if they were member 
functions of the char[].

#alias char[] string;
#size_t find_last_of( string _this, dchar ch ) { ... }

#string str = "whatever";
#size_t p = str.find_last_of('e');

This must be my favorite feature in D!

L.
Jun 26 2006