Switching to HTML5, take two

[Aryeh Gregor]

Back in July, we had a lengthy discussion about the possibility of
switching to HTML5:

http://lists.wikimedia.org/pipermail/wikitech-l/2009-July/043865.html

The conclusion appeared to be that switching to HTML5 was a good idea.
 As of July 12 (r53142), MediaWiki has been outputting HTML5
continuously as the default:

http://www.mediawiki.org/wiki/Special:Code/MediaWiki/53142

There have been only a few minor problems reported by various trunk
users, all of which were quickly fixed.  The current state of HTML5
support is discussed here:

http://www.mediawiki.org/wiki/HTML5

When we had our last scap six weeks ago, $wgHtml5 was disabled on
Wikimedia to reduce the number of variables to keep track of, since a
huge amount of code was changing at once.  Although we're currently in
the middle of some organizational issues due to Brion's resignation, I
think it would be safe to experimentally turn on $wgHtml5 and see what
happens.  If any problems are reported that can't be fixed
immediately, of course, it could be turned off again, but I think this
is quite unlikely.  I'll be available for the next few weeks to work
on any issues that arise.  Would this be okay?

_______________________________________________
Wikitech-l mailing list
Wikitech-l@...
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Max Semenik]

On 04.11.2009, 19:28 Aryeh wrote:

> Back in July, we had a lengthy discussion about the possibility of
> switching to HTML5:

. . .

> Would this be okay?

Fine by me as a bot developer, the lack of XML well-formedness will
not break AWB. Those poor souls who still screen-scrape should expect
all kinds of breakages anyway. The API is mature and was there for
enough time for everyone who cares to migrate.

--
Best regards,
  Max Semenik ([[User:MaxSem]])


_______________________________________________
Wikitech-l mailing list
Wikitech-l@...
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Platonides]

Max Semenik wrote:
> Those poor souls who still screen-scrape should expect
> all kinds of breakages anyway.

That's no reason not to warn them ahead of time. Crossposting the notice
to wikibots-l and pywikipedia-l.

A change like wgHtml5=true seems a good time to change the legacy code
to use the api.

Please report anything doable with screen-scrapping but not with
mediawiki api.


Aryeh Gregor writes:

> Back in July, we had a lengthy discussion about the possibility of
> switching to HTML5:
>
> http://lists.wikimedia.org/pipermail/wikitech-l/2009-July/043865.html
>
> The conclusion appeared to be that switching to HTML5 was a good idea.
>  As of July 12 (r53142), MediaWiki has been outputting HTML5
> continuously as the default:
>
> http://www.mediawiki.org/wiki/Special:Code/MediaWiki/53142
>
> There have been only a few minor problems reported by various trunk
> users, all of which were quickly fixed.  The current state of HTML5
> support is discussed here:
>
> http://www.mediawiki.org/wiki/HTML5
>
> When we had our last scap six weeks ago, $wgHtml5 was disabled on
> Wikimedia to reduce the number of variables to keep track of, since a
> huge amount of code was changing at once.  Although we're currently in
> the middle of some organizational issues due to Brion's resignation, I
> think it would be safe to experimentally turn on $wgHtml5 and see what
> happens.  If any problems are reported that can't be fixed
> immediately, of course, it could be turned off again, but I think this
> is quite unlikely.  I'll be available for the next few weeks to work
> on any issues that arise.  Would this be okay?

_______________________________________________
Wikitech-l mailing list
Wikitech-l@...
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Aryeh Gregor]

On Thu, Nov 5, 2009 at 4:01 PM, Platonides <Platonides@...> wrote:
> That's no reason not to warn them ahead of time. Crossposting the notice
> to wikibots-l and pywikipedia-l.
>
> A change like wgHtml5=true seems a good time to change the legacy code
> to use the api.
>
> Please report anything doable with screen-scrapping but not with
> mediawiki api.

I don't expect any bots to break, even if they screen-scrape.  The
actual changes to HTML output are minimal -- a different doctype, a
few useless elements and attributes dropped.  Unless your script
relies on the fact that all <script> elements have
type="text/javascript" or something similarly crazy, you should be
fine even if you screen-scrape.

Switching $wgWellFormedXml to false would certainly break
screen-scrapers that rely on XML parsing libraries.  However, I don't
propose we do this right now -- I'm only suggesting we output HTML5,
not cease outputting well-formed XML.  Eventually, perhaps, we should
consider dropping XML output, especially once HTML5 parsing libraries
are more readily available, but not in the foreseeable future.

The only breakage I can foresee would be caused by the switch from
"almost standards" to "standards" mode.  But the difference is pretty
minor (far more minor than quirks vs. almost standards), and I don't
expect anything to break seriously, or in a way that can't be quickly
fixed.  No trunk users (TranslateWiki, etc.) have reported any
outstanding problems that I know of.

_______________________________________________
Wikitech-l mailing list
Wikitech-l@...
https://lists.wikimedia.org/mailman/listinfo/wikitech-l