May 21, 22, 23 TC39 Meeting Notes

Kevin Smith zenparsing at gmail.com
Tue Jun 4 06:22:02 PDT 2013


I cannot agree with Andreas more.  In my opinion, module registrations
should not be used as an authoring format, but only as a target for tooling
(specifically multiplexers).

In any case, I don't see how there is _true_ consensus around this issue at
this time, nor do I see that anything is set in stone quite yet.  In my
view, the only thing holding us back is the lack of lexical modules.  It
appears to me that:

- Everyone involved wants lexical modules at _some_ point
- There are some participants that consider the lack of lexical modules a
"fundamental flaw".
- The semantics of lexical modules are not really in dispute

I have created an initial implementation of a package bundler using lexical
modules here:

https://github.com/zenparsing/js-bundle

An example output of the tool is here:

https://gist.github.com/zenparsing/5705769

I could fairly easily change the tool so that it uses global registrations,
instead of lexical modules.  I'd have to use arbitrary random strings as
module names to minimize the chance of collision with some other valid
file, but it's possible.  But note:

Using global module registrations for bundling means that it is not
possible to hide internal modules from external clients!

I consider the inability to create high integrity packages to be a
fundamental flaw.  The simple addition of lexical modules to the current
design will correct this flaw.

{ Kevin }
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/es-discuss/attachments/20130604/caf8c345/attachment-0001.html>


More information about the es-discuss mailing list