[Laszlo-dev] Need uniform naming scheme for classes

[Henry Minsky]

We'll  still put a pointer to the class as an entry in the global "lz",  I.e.,

<class name="foo">

gets a pointer as lz.foo, which is the documented way to access them.
So the actual classname of a userclass  is going not going to be
advertised to the users.
 It seems like $lzc$foo would be fine for a user-defined classname. We
wouldn't be renaming
and of the existing LFC classes, right?

On Tue, Mar 11, 2008 at 8:13 AM, P T Withington <ptw at pobox.com> wrote:
> I need to be able to name both user classes and built-in (LFC) classes
>  from their corresponding tag name.  In JS1 we can get away with
>  (effectively) saying:
>
>  <class name="usertag" extends="anothertag"> ...
>
>  maps into:
>
>  class <unique prefix> + usertag extends ClassMap[anothertag] { ...
>
>  but that is clearly not legal JS2.  [We need the unique prefix to keep
>  the class from shadowing any runtime class, e.g., the user class
>  `window` would shadow DHTML `window` otherwise.  In JS2 we can use a
>  namespace for this purpose.  We could extend the JS1 translator to
>  fake namespaces, but I don't want to take on that task at this point.]
>
>  In the LFC there is a mapping from tag name to class name `view`
>  becomes `LzView`.  I'm not completely opposed to continuing that
>  practice, but I would prefer using our new convention of `view`
>  becomes `$lzc$view` because:  a) the leading `$` indicates that this
>  is not a public name, b) the interior `$` serves to delimit the prefix
>  from the tagname making it easier to unparse, and c) it is closer to a
>  namespace syntax.
>
>  Comments?  Is this a useful change to make, or am I just making work
>  where it is not necessary?
>



-- 
Henry Minsky
Software Architect
hminsky at laszlosystems.com