Re: Namespace Qualifiers

new topic     » goto parent     » topic index » view thread      » older message » newer message

On Sun, 9 Mar 2003 10:51:03 +1100, Derek Parnell
<ddparnell at bigpond.com> wrote:

>You can't. Euphoria's namespace concept does not help us in this =
situation.
>
>I would reason that because win32lib does NOT include win_misc.ew, =
either
>explicitly or implicitly, that the 'or_all' that win32lib is referring =
to is
>the one in tk_maths.e, which win32lib DOES include. Why Euphoria thinks =
that
>win32lib might be trying to refer to a version of a routine that it had =
no
>knowledge of is weird.
>
>What I need to do is to put namespace qualifiers on every routine that
>win32lib references. There must be only a few thousand of those....

So what is the "proper" solution then?

a) change std Eu to automagically resolve duplicate globals based on=20
     include level; an included file referencing an included global it
     has itself included clearly means that one and no other.

b) get smart; the or_all routine in tk_maths.e and the one in=20
     win_misc.ew are nothing if not identical; they should resolve to=20
     the exact same bytecode. Better than moaning about a conflict,=20
     the interpreter should just take the first; the binder/translator
     optimise the second out of existence.

c) extend namespace references somehow at the top level to state that
I=20
     mean the "or_all" sub-included by file x rather than the one=20
     sub-included by file y?

Pete

new topic     » goto parent     » topic index » view thread      » older message » newer message

Search



Quick Links

User menu

Not signed in.

Misc Menu