W3C home > Mailing lists > Public > public-mobileok-checker@w3.org > October 2007

Re: mobileOK reference implementation alpha release ready

From: Dominique Hazael-Massieux <dom@w3.org>
Date: Thu, 18 Oct 2007 11:03:12 +0200
To: Sean Owen <srowen@google.com>
Cc: public-mobileok-checker <public-mobileok-checker@w3.org>
Message-Id: <1192698192.24095.9.camel@cumulustier>

Le mercredi 17 octobre 2007 à 20:07 -0400, Sean Owen a écrit :
> Thanks Dom, I fixed this and a few other related issues that crop up
> when you turn this loose on a regular web page. I guess I didn't
> really try that.

Great, it works much more reliably now indeed!

> Two observations:
> 
> - the checker noted that http://www.w3c.or.kr/ currently does not resolve. :)

Yeah, that happens on a regular basis for some reasons... Nothing I can
about it, unfortunately.

> - It takes a long time to retrieve all those linked resources in this
> case. After, what, 30 links, should we give up on the assumption that
> it's going to fail a bunch of tests because it probably isn't mobile?

We need some sort of parametrization for this indeed; I don't think we
should have a hardcoded limit (given that the number of links you want
to set as a limit might depend on the context in which you use the
library).

That said, the number of links shouldn't have such a bad impact on the
library; the current BP checker used to have a big problem with that
until I made it use threads, as well as a making sure a given URI would
only be checked once; from what I remember the library does this too,
but that may be worth a check.

> I'm concerned that this library's strongest talent seems to be load
> generation at the moment.

:)

I'm actually very impressed with what the library already does (beyond
generating load); it sure does need some refinement before I can make an
official Web interface on top of it, though.

A few additional remarks:
 * some FAIL results (at least the ones from CONTENT_FORMAT_SUPPORT)
generate escaped XML to give the error location and the details about
the error, e.g.:
        <result name="CONTENT_FORMAT_SUPPORT_AND_VALID_MARKUP-4"
        outcome="FAIL">
                 <info>FAIL: If the document is an HTML document and it
        fails to validat
        e according to its given DOCTYPE, FAIL</info>
                 <code>&lt;error&gt;&lt;info&gt;The content of element
        type "p" must mat
        ch "(br|span|em|strong|dfn|code|samp|kbd|var|cite|abbr|acronym|
        q|i|b|big|small|a
        |img|object|input|select|textarea|
        label)".&lt;/info&gt;&lt;position type="LINE_C
        OLUMN"&gt;&lt;line&gt;50&lt;/line&gt;&lt;column&gt;5&lt;/column&gt;&lt;/position
        &gt;&lt;/error&gt;</code>
              </result>
I assume this is a bug.
 * some error messages seem to be undefined:
   e.g. "message id STYLE_SHEETS_USE-5 in file messages.properties.xml
not defined."
(I'll have a look at fixing that particular one, but I wonder if there
is a well-defined way to check that each message id has a corresponding
entry in the properties file)

Dom
Received on Thursday, 18 October 2007 09:04:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:13:04 GMT