[QA] I had to delete the ProveIt gadget from beta labs

Matthew Flaschen mflaschen at wikimedia.org
Sat Oct 26 00:25:52 UTC 2013


On 10/25/2013 12:17 PM, Chris McMahon wrote:
> * because of the way Selenium/WebDriver works in Chrome, some Chrome
> tests would fail due to the way that ProveIt lies on top of all of the
> other page controls.  ChromeDriver thinks that e.g. the 'Save' button in
> the wikitext editor cannot be clicked, because it lies "behind" the
> ProveIt controls.

Yeah, this is annoying, but I'm not sure what to do about it.

> * ProveIt code being loaded into VisualEditor.  This is probably a VE
> bug.  See the screen shot
> https://bugzilla.wikimedia.org/show_bug.cgi?id=55801

Loading ProveIt in the actual VE edit page (e.g. 
http://en.wikipedia.org/wiki/Earth?veaction=edit) is undesirable.  But 
it takes some work to prevent, since gadgets have no say in which pages 
they're loaded on.  The only option is to make a loader script, but that 
adds a round-trip, and I'm not sure if it's worth it.

However, this seems to be about all gadgets (not just ProveIt) being 
loaded into the iframes.  Indeed, this does not seem to be useful, since 
the gadget on the top-level article can reach into the frames already. 
I replied noting this.

> * The javascript and css would be loaded erroneously in Internet
> Explorer, sometimes causing the wikitext editor to fail to load.  I
> would see the ProveIt styling on the upper left of the page in this case.
>    See the screen shots at
> https://bugzilla.wikimedia.org/show_bug.cgi?id=56117

The JS error you found is actually in RefToolbarLegacy.  It is possible 
that RL is loading ProveIt when it shouldn't, though I have seen no 
other sign of such a bug so far.  I've replied in more depth at the bug.

> So testing ProveIt served its purpose-- we found some good bugs by doing
> it.  Now to get the bugs fixed.

If you find something you believe is a bug in ProveIt itself, please 
file at https://code.google.com/p/proveit-js/issues/list .

Thanks,

Matt Flaschen




More information about the QA mailing list