jacob.farber.work at gmail.com
Wed Jan 21 19:00:12 UTC 2009
Thanks very much for the feedback- I have 2 main points inline:
I have attached a screen shot I took a while back when I ran into a problem
while testing the robot automation. The image shows the same page opened in
FF2 and FF3. Notice the difference in size of the portlets. The browsers
both had the same size.
Yes, there is a huge difference between browsers, but I can only assume what
would happen is there would be one master image per browser + OS combo (so
FF3 on XP would have one master image, and FF3 OSX would have another, and
neither would be expected to look exactly the same).
I imagine with this method that would include creating new screen shots that
are being compared against. I'm wary of writing tests that compare against
Im a little worried about maintaining a bunch of master images too, esp.
when the system is so young that things have not stabilized yet. I wonder if
I could automate screenshot taking across multiple browsers/platforms....
*Some futher thoughts:*
One other way of testing that would be to manually view a test page in each
browser + OS combo, and inspect it for anomalies. This too seems tedious and
time consuming, and riddled with human error.
Another way would be to programmatically inspect the computed style values
of every node on a test page in every browser + OS combo. This seems the
most automatic, yet has the potential to completely miss rendering errors
(something could have the correct computed style yet still be completely
Any other thoughts? This discussion is very very helpful
On Wed, Jan 21, 2009 at 11:44 AM, <michelle.dsouza at utoronto.ca> wrote:
> I haven't looked at this very closely but it sounds like a very high
> maintenance way of doing testing. Every time I've built user interface tests
> that relied on what things look like the maintenance cost has been too high
> to sustain them. Essentially, minor code tweaks results in hours of updating
> test cases. I imagine with this method that would include creating new
> screen shots that are being compared against. I'm wary of writing tests that
> compare against static images.
> Quoting Jacob Farber <jacob.farber.work at gmail.com>:
> Has anyone ever done automated testing for CSS before? If so, have you
>> of http://github.com/garethr/css-test/tree/master or comparing a rendered
>> webpage image to an master image? I would be interested to hear about any
>> experiences you may have had with this technique.
>> Jacob Farber
>> University of Toronto - ATRC
>> Tel: (416) 946-3002
University of Toronto - ATRC
Tel: (416) 946-3002
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fluid-work