I'd like to propose a change in how we integrate new
languages. Currently, the process is:
- someone write the BSFEngine implementation
- get it integrated to Apache BSF CVS tree
- update the languages.props file to point to it
The main problem with this is that this ties language updates
to BSF version updates. Also, the people who really understand
how the engine works are the language implementors and hence
one of them has to be in BSF commit land to make this work.
An alternate is to encourage language implementors to include
a BSF engine impl in their distribution. For Java-based langs
this is trivial of course. So that decouples the language
stuff from BSF and the language community is free to update
the engine whenever they update the language. If BSF does
change, then of course they need to update too. Given that BSF
is likely to be evolving slower than the language impls (IMO)
this makes sense.
The only thing we'd do is add an entry into language.props.
Once we get a dynamic discovery mechanism in we won't even
need that. Then a BSF user that wishes to use language "foo"
simply needs to know that name "foo" that the language authors
picked for their stuff and all else is automatic.
What do you think?
Sanjiva.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
