[Wikitext-l] Visual Editor / parser status update meeting notes 2011-10-26
brion at pobox.com
Wed Oct 26 19:39:56 UTC 2011
Etherpad'ed notes from today's editor/parser status meeting:
* Visual Editor team
* Parser team
** Gabriel Wicke (skype)
* Managy people
== Status check ==
How are we going to
- Write a parser
- Integrate it with the visual editor
- Deploy the visual editor
- Basic parser in place
- Produces intermediate JSON object tree
- Doesn't yet produce Wikidom as it's been defined
- Considering if Wikidom should be changed
- Still working on markup support (mixed HTML and Wikitext)
- Using PEG for part of the parsing process (primarily tokenizing)
- Get something working in the wiki environment so people can play with
- Ideas for places to integrate first
- New page creation - Potential Use Case
== Decemberish test deployment? ==
Distinct stages that could be deployed on sites for testing:
* Visual editor in the wiki -> lets you edit but nothing else
** ''too early to advertise for anything but interaction testing''
** just need to actually plug it in
* Visual editor in the wiki and lets you save
** ''_could_ start to use as opt-in test for new pages''
** add API module to save, should be easy-ish
* Visual editor in the wiki and lets you load and save.
** ''this would be ideal place to start advertising for testers''
** combining the parser and reconciling the formats
== Visual Editor side ==
VE is in midst of core redo to make undo etc working (internal
representation changes)-- still stabilizing!
VE frontend todos for page saving:
* reintegrate serializers (in progress)
* undo/redo/copy/paste working (mostly there)
** '''Trevor and Inez are already working on this, but it needs to be
finished before we can integrate saving.'''*
Parser/VE integration todos (for page loading):
* parser & wikidom specs need to be resynchronized
** possibly needs a translator step
* support for more structures (lists etc)
** '''This is big work and needs a lot of coordination between VE & Parser
** brion & gabriel: can start working with their existing spec -- let's at
least get that working
*** then we'll need to start figuring out if we need to do major changes
based on issues
*** parser & VE may need to evolve together for now!
Lots more front-end work needed in VE ecosystem; we expect to bring on
another hire in that area.
"WikiDom will be the fence between these two neighbors [VE & Parser groups]"
== Parser side ==
*Immediate parser work:*
* get it producing output compatible with VE's current wikidom input
** we do expect to have to evolve this over time! but let's get it together
with the VE first
* make sure there's a clean JS API for the parser so the VE can call it
*Next stages parser work:*
* start making sure tables, templates are all working as expected (fixup
** at least properly nested templates..
* iterate as necessary
* make sure auto tests are running
** round-trip tests on wikipedia corpus already available
** need to test HTML rendering against MW's current internal parser
*Later stages parser work: (after December)*
* depending on what we've seen from auto test...
* ... redoing the PEG?
* ... redoing the wikidom?
* ... redoing how/when templates are expanded?
* make sure all structures really supported!
* add proper fix-ups similar to what tidy does right now
== Upcoming work ==
* Mediawiki extension for Visual Editor: Architecture is to be driven by
Trevor; Integration w Parser
* RL2 support
* Internationalization support
* Using underscore.js for js based parser functionality in the same
extension as VEd
** replace jQuery bits like $.each; underscore should be cleaner and is
easier to use in node for batch tests
* Unknown markup indicator
** at least black-box markers for extension hooks and such
* Trevor will check up with Howie
** various VE stuff on its way
* getting VE and parser into a common extension (separate code modules
loaded by it)
* parser integration may not make it by December, but we'll see what's ready
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Wikitext-l