The shared modules for the calendar Mozmill-Tests use some modules, which are not available in the comm-central repository. These can be replaced by others used by Thunderbird in Mozmill-Testing thus enabling calendar tests to run as packaged tests.

Comment on attachment 8434421 [details] [diff] [review] mozmillSharedModules.diff Review of attachment 8434421 [details] [diff] [review]: ----------------------------------------------------------------- r=philipp if you can present a green try-run with this applied :-) ::: calendar/test/mozmill/shared-modules/test-calendar-utils.js @@ +7,4 @@ > > +Cu.import("resource://gre/modules/Services.jsm"); > +var os = {}; Cu.import('resource://mozmill/stdlib/os.js', os); > +var frame = {}; Cu.import('resource://mozmill/modules/frame.js', frame); I believe you can also write this as: var frame = Cu.import('...', {}); @@ +59,5 @@ > + */ > +function getCalProperty(bundleName, propId) { > + var props = Services.strings.createBundle( > + "chrome://calendar/locale/" + bundleName + ".properties"); > + return props.GetStringFromName(propId); Can't you use cal.calGetString for this?

> I believe you can also write this as: > var frame = Cu.import('...', {}); Yes, that works. And I present to you: A working try-run: https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=56d4151e98ed The failures on linux are not in our tests... > Can't you use cal.calGetString for this? There had been problems loading the calendar-modules in the tests. Meanwhile I believe it works. For now I can leave this function out, because I will not need it until the timezone-tests are fixed. There I can try to use calGetString. With this changed, I am going to push soon.

updated patch per the last comments r=philipp

Comment on attachment 8442871 [details] [diff] [review] patch V2 r=philipp There are changes to be made to this patch for mac compatibility as I just found out. Therefore this is obsolete and checkin-needed cancelled.

This is a patch with updated "setData" function. I changed all the blocks containing "if (!mac)" statements to be platform independant. For that, I took out combinations of "a + ctrlKey" since the ctrlKey does not work on Mac. Instead, now all these parts contain ".select()" on the text-fields which also causes the whole text to be selected. Surely this does not imitate the user interaction exactly, but since we are not testing for the behaviour of textboxes, I think this is acceptable. I could not test the patch, because the builds are busted at the moment. Philip, please test this when you have a working building env. Unfortunately there are no tests working at the moment, which use this function, but we can make sure the module works overall.

Comment on attachment 8443925 [details] [diff] [review] patch V2.1 Review of attachment 8443925 [details] [diff] [review]: ----------------------------------------------------------------- I ran our three basic tests with only this patch applied locally and it seems to work. To be sure, I've started a try run too. I hope the recent bustages don't hinder the builds, I haven't updated in a while so we might be safe. https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=562436af1f0f Code looks fine, r=philipp If the try build passes, go ahead and push.

New try run for this bug and bug 980515 now that c-c is green again: https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=2d130d00f10e

Does this have importance in the effort to move to ical.js?

I don't think it does. This would be nice cleanup for the mozmill tests though. I'm making some updates in bug 1303626 so this patch may need some updates.

Debitrotted pacth after Bug 1297425

I guess, this patch was intended for bug 1303626?

Uh, yes. I was too tired and too quick yesterday...