634907 - need way for mozmill tests to start up without accounts

Status: RESOLVED FIXED

(Thunderbird :: Testing Infrastructure, defect)

Description

[David :Bienvenu]

Attachment: sid's working patch

For testing the setup instrumentation code, it would be very helpful to have a way for mozmill tests to start up without the accounts pre-setup, so we can simulate initial run after download. 

Sid has already come up with a patch for this, which works great for my purposes. If you add a wrapper.py file containing NO_ACCOUNTS = True to your test directory, your profile won't have accounts pre-setup at load.

Comment 1

[Mark Banner (:standard8)]

So we already have an option to pre-load a prefs.js file (see migration-from-tb2), I'm not convinced that its necessary to add a no accounts option considering you can just create your own prefs.js for the directory.

Comment 2

[David :Bienvenu]

(In reply to comment #1)
> So we already have an option to pre-load a prefs.js file (see
> migration-from-tb2), I'm not convinced that its necessary to add a no accounts
> option considering you can just create your own prefs.js for the directory.

My understanding is that the pre-loaded custom prefs js file is additive, not replacing the existing prefs.js file, and in any case, the existing prefs.js file is loaded last, so whatever is in there wins. And that's the file that has the accounts set up.

Comment 3

[David :Bienvenu]

Standard8 and I talked about this - he was suggesting overwriting the prefs.js file with my own prefs.js file w/o the account manager settings. I believe I convinced him that was a bit fragile, because every time someone changes the prefs in runtest.py, they'd need to make a similar change to my test.

Comment 4

[Rain [:rain1] (inactive)]

Yeah, that sounds right.

Comment 5

[David :Bienvenu]

so who should review this? Standard8?

Comment 6

[Mark Banner (:standard8)]

Comment on attachment 513134 [details] [diff] [review]
sid's working patch

>+        # Do not allow fonts to be upgraded
>+        'mail.font.windows.version': 1
...
>-        # Do not allow fonts to be upgraded
>-        'mail.font.windows.version': 2

Is this change intentional, if so why?

Comment 7

[Rain [:rain1] (inactive)]

Good catch -- no, it isn't intentional. I think it was still 1 when I sent bienvenu the patch, but I changed it to 2 later in the tree. When the patch was rebased the deletion was updated but the addition wasn't.

Comment 8

[Mark Banner (:standard8)]

Comment on attachment 513134 [details] [diff] [review]
sid's working patch

r=Standard8 with the previously mentioned item fixed.

sid, could you land this?

Comment 9

[Rain [:rain1] (inactive)]

https://hg.mozilla.org/comm-central/rev/4c46f52d847d