780227 - Support different markup languages

Status: RESOLVED FIXED

(developer.mozilla.org Graveyard :: Editing, defect)

Description

[Benoit Jacob [:bjacob] (mostly away)]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/17.0 Firefox/17.0
Build ID: 20120729030516

Steps to reproduce:

Try to edit a page on MDN.


Actual results:

Only options offered are WYSIWYG or manual editing of HTML.


Expected results:

Possibility of editing in a friendly markup language as in every other developer site that I've seen:
 - either mediawiki
 - or markdown

Comment 1

[Jeff Hammel]

... or restructured text or ...

See also https://bugzilla.mozilla.org/show_bug.cgi?id=841973 for an out of band solution

Either in-band or out-of-band, a general solution is as follows:

User->server:
- user enters {{text-to-HTML}} syntax (where applicable, see Server->user)
- {{text-to-HTML}} is rendered to HTML

Server->user:
- HTML is converted to {{text-to-HTML}} syntax where possible
- since the HTML core may contain HTML not convertible to {{{text-to-HTML}} format, the remaining portions are presented as raw HTML; if {{text-to-HTML}} syntax supports embedded HTML, that is used; otherwise e.g. the multiple parts are presented separately (UI/UX challenge)

{{text-to-HTML}} apropos commment 0 is either markdown or mediawiki

Obviously this is a napkin sketch, not a spec. I would guess in implementation, several challenges will be encountered.  I'm (very) happy to hash this out (here or on bug 841973 , as appropriate).

Comment 2

[Jeff Hammel]

Some HTML->mediawiki links:

* http://www.mediawiki.org/wiki/Manual:Importing_external_content#Converting_content_from_HTML_text_file
* http://search.cpan.org/~diberri/HTML-WikiConverter-0.68/ (perl)
* http://code.google.com/p/gwtwiki/wiki/HTML2Mediawiki (java)
* https://migrations.atlassian.net/wiki/display/UWC/Universal+Wiki+Converter (Java)


See also: http://techwiki.openstructs.org/index.php/Wiki_converters

Comment 3

[Les Orchard [:lorchard]]

The Kuma wiki supports a liberal subset of HTML in round-trip editing with a psuedo-WYSIWYG editor. This design decision gives us flexibility in doc structure, but I think rules out things like MediaWiki markup.

I can't think a non-lossy process to support round-trip editing with both HTML+WYSIWYG and MediaWiki at once. Using the HTML WYSIWYG will add attributes and markup that MediaWiki doesn't support. Conversely, using MediaWiki will destroy attributes and markup added in the WYSIWYG.

Replace "MediaWiki" with "Markdown", "Textile", "reStructuredText", and you have pretty much the same basic issue.

As an alternative, we've been tossing around ideas for external doc sources (ie. from github, hg, etc) and started bug 861981. The notion of different markup languages could be addressed there, at least, insofar as a discrete bundle of docs could settle on a non-HTML markup type.

Comment 4

[James Socol [:jsocol, :james]]

We have a corpus of 10s of thousands of documents in HTML. The site also *documents* HTML. Is there any data that we're actually losing contributions to the wiki because of the editor or HTML?

Supporting multiple markups seems like a massive engineering effort and unless there's massive problem we're solving, I think our time is much better spent elsewhere.

There is bug 861981, which is about handling documentation from other sources in other formats, but not within the wiki itself.