W3C home > Mailing lists > Public > public-mwts@w3.org > February 2009

http://dabase.com/guidelines.html

From: Kai Hendry <hendry@iki.fi>
Date: Tue, 3 Feb 2009 16:53:21 +0000
Message-ID: <b24851260902030853m1e53ce9duc7483bcb8774c418@mail.gmail.com>
To: public-mwts <public-mwts@w3.org>

Firstly I just want to say that the guidelines document is looking
much better. Great work guys.

I copied the http://dev.w3.org/2008/dev-ind-testing/guidelines.html to
my local server and changed some dl to ol and fixed a typo. I prefer
ol over ul as people can say:
"I have a problem with point 2 of 'Extension capabilities'" instead
of, the second bullet. No biggie. Also you can nicely see a history of
my changes:
http://git.webconverger.org/?p=faq.git;a=history;f=guidelines.html

If you like them I can change the W3C version.


As for other ideas:
1 encourage short test urls
2 encourage people to lint/validate test code, like
http://wiki.whatwg.org/wiki/IDE
3 encourage javascript, css and html to be separated to make validation easier


Really didn't like point 3 of Target devices. What test for e.g.
requires server side adaption?

I didn't agree with "When using colors to express the result of the
test, convey the result also with text.". I think colours are enough.


As I mentioned during the call, "memory limitations" could be reworded
to be something to be aware of. I really doubt someone is mad enough
to write a test that would accidentally go over the limit here. I
would simply say keep your line count low.

nox:/srv/www/example.webvm.net/template% wc -l *
 20 index.html
 14 test.js
 34 total

<50 lines of code or something like that.




I thought it might be worth mentioning emulators. As we all know,
nothing beats testing on the actual device.

Next I was going to perhaps mention using the lowest common
denominator of <body onload="". Or making the buttons on interactive
tests bigger. Or perhaps a tip about recording the user agent.

I was thinking. Is there a good resource explaining some testing
terminology that we might use? Smoke test, composite, false positive.
Something of theoretical/academic quality. Perhaps
http://www.w3.org/Style/CSS/Test/guidelines.html#kinds is sufficient.


Thanks guys,
Received on Tuesday, 3 February 2009 16:55:03 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:08:02 UTC