www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - new DIP14: Package import path association

reply "Martin Nowak" <dawg dawgfoto.de> writes:
http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14
Oct 18 2011
next sibling parent reply Jacob Carlborg <doob me.com> writes:
On 2011-10-18 09:38, Martin Nowak wrote:

 http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14


What happens if you import a the same file twice using different paths, 
i.e. relative and absolute path?

Will the imported files need a module declaration or will the compiler 
just care about the path?

-- 
/Jacob Carlborg
Oct 18 2011
next sibling parent Gor Gyolchanyan <gor.f.gyolchanyan gmail.com> writes:
I think the best thing to do would be to maintain a module[path]
mapping, where more then one path can map to a single module.
The by-path import declarations would add items to the mapping and the
compiler would extract the list of modules from it.

On Tue, Oct 18, 2011 at 3:23 PM, Jacob Carlborg <doob me.com> wrote:

 On 2011-10-18 09:38, Martin Nowak wrote:

 http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14


 What happens if you import a the same file twice using different paths, i.e.
 relative and absolute path?

 Will the imported files need a module declaration or will the compiler just
 care about the path?

 --
 /Jacob Carlborg
Oct 18 2011
prev sibling parent reply "Martin Nowak" <dawg dawgfoto.de> writes:
On Tue, 18 Oct 2011 13:23:02 +0200, Jacob Carlborg <doob me.com> wrote:


 On 2011-10-18 09:38, Martin Nowak wrote:

 http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14


 What happens if you import a the same file twice using different paths,  
 i.e. relative and absolute path?

 Will the imported files need a module declaration or will the compiler  
 just care about the path?


The module name is determined by the module declaration or implicitly by  
the basename of the file.
Module collisions are already detected based on the module name.
You can actually import two identical modules through different  
files/symlinks as long
as they don't have a module declaration.
This does not differ from the current behavior.

martin
Oct 18 2011
parent Jacob Carlborg <doob me.com> writes:
On 2011-10-18 13:48, Martin Nowak wrote:

 On Tue, 18 Oct 2011 13:23:02 +0200, Jacob Carlborg <doob me.com> wrote:


 On 2011-10-18 09:38, Martin Nowak wrote:

 http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14


 What happens if you import a the same file twice using different
 paths, i.e. relative and absolute path?

 Will the imported files need a module declaration or will the compiler
 just care about the path?


 The module name is determined by the module declaration or implicitly by
 the basename of the file.
 Module collisions are already detected based on the module name.
 You can actually import two identical modules through different
 files/symlinks as long
 as they don't have a module declaration.
 This does not differ from the current behavior.

 martin


What happens then, to module constructors and similar? What concerns me 
is that something similar to Ruby 1.8 could happen.

In Ruby when the compiler encounters a "require" it will load the file 
specified in the require call. Meaning any top level code will be 
executed. If you require the same file again, i.e. from another file it 
will not execute the required file again because it's already loaded.

The problem with this is that the load path which stores all loaded ruby 
files does not sore the full path name of a ruby file, it stores what 
every you called "require" with.

If you then require the same file using two different paths, i.e. 
relative and full path, the required file will be executed again, 
possibly resulting in warnings like:

warning: already initialized constant FOO

This is fixed in Ruby 1.9 by always storing full paths to required files 
in the load path.

-- 
/Jacob Carlborg
Oct 18 2011
prev sibling parent Don <nospam nospam.com> writes:
On 18.10.2011 09:38, Martin Nowak wrote:

 http://prowiki.org/wiki4d/wiki.cgi?LanguageDevel/DIPs/DIP14


Your proposal seems to have an implicit "Posix file path syntax is 
used", but that opens a can of worms.
Up until now, there has been a policy of ensuring that OS-specific paths 
are kept out of the language.
Oct 18 2011