The visual editor looking the same as the view is just one idea, and in some
ways we know the editor will look a little different, so it's probably not
that good of a use case for this kind of testing.
I was more thinking about interactions, such as you click on a link, some
options popup below it, you select one, you get a dialog, etc. This kind of
testing is best for testing interactions, and it would really shine in this,
and other similar cases.
- Trevor
On Mon, Sep 19, 2011 at 3:45 PM, Markus Glaser <glaser(a)hallowelt.biz> wrote:
I've been
aware of this tool for quite a while, and shown it to some
other devs around
here. I think it's awesome, but I have not
had a need for it yet. I think
the visual editor
may present some cases where this makes sense
Just being curious, can you elaborate on that? From my experience (I tried
TinyMCE on MediaWiki ;), would it not be easier to compare the code a visual
editor produces with the result of the preview mode? I assume it could
easier be done by comparing the DOM branches, using QUnit or Selenium
instead of images.
- but generally it seems the most useful for
writing tests that involved
taking several input
actions and expecting a consistent result.
I agree. I thought more about testing skin layout, e.g. divs not being
rendered in the right place in some browsers. For complex interactions, I'd
still prefer Selenium or other "non-optical" tools.
-- Markus (mglaser)
On Mon, Sep 19, 2011 at 7:30 AM, Markus Glaser <glaser(a)hallowelt.biz>
wrote:
Hi,
while I don't like the idea of introducing more and more testing
tools, I can still see an interesting use case here: as of now, we
have no way to test whether a given layout (HTML, JS, CSS) is really
rendered the way we want it to be, since both Selenium and QUnit make
their tests based on DOM, right? Sikuli on the other hand seems to be
based on screenshots and here we could detect broken layout. There is
also some kind of similarity algorithm (which I hope is configurable)
so that one test could be used in different browsers even if the
rendering is not
identical to the pixel.
The question is, do we have the need for testing screen layout?
Cheers,
Markus
P.S.: CCing wikitech, since this might be of broader interest.
-----Ursprüngliche Nachricht-----
Von: Sumana Harihareswara [mailto:sumanah@wikimedia.org]
Gesendet: Dienstag, 30. August 2011 14:02
An: Markus Glaser; Chad Horohoe; Timo Tijhof
Betreff: automated testing with Sikuli?
http://sikuli.org/
Have any of you run across Sikuli before? Just wanted to point it out
to you. It might face the same problems as Selenium, though.
--
Sumana Harihareswara
Volunteer Development Coordinator
Wikimedia Foundation
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l