Followup: modularity, WebExtensions, and going faster

[Richard Newman]

Excellent summary, mconley. (And excellent thoughts besides.)

I think most of the pushback on this list is to the ideas of:

1. Splitting up *current* pieces (emphasis on the plural) of Firefox,
somehow making it easier to work on them. That presupposes that our modules
are currently close to correct.
2. Shipping these pieces separately, somehow making it easier to move
faster. That presupposes that the workaday troubles of contributing to
Firefox are more important than conceptual load.
3. WebExtensions being a fashionable panacea. To me that sounds like B2G
WebFoo.

Without these, one reduces the argument to "better modularity would be
nice, and it might help to ship some things at different rates when
suitable" — and I think we can agree on both of those!


Or, to take rnewman's angle here, is there an opportunity here to draw some
> new boundary lines and define a new (but small) WebExtension Thing that
> combines capabilities from several of our current modules for great good?


I'm hoping to end up with Datomish being one such cross-cutting feature. It
has a very small defined API surface, is fully asynchronous, and is
designed for cross-module use. At some point I'm planning to see if I can
implement the stock Places (and Satchel, and…?) APIs on top. At the very
least it'll be an interesting performance test!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/firefox-dev/attachments/20161012/394e8215/attachment.html>