703211 - number of recipients in header pane (litmus tests #12638 and #13686)

Status: NEW

(Thunderbird :: Folder and Message Lists, defect)

Description

[Eric Moore]

I'm running "20: Check number of recipient in header pane" in the "navigating and displaying messages" Litmus test using Earlybird (10.0a2) while running in safe mode using a IMAP account. mailnews.headers.show_n_lines_before_more was already set to 1.

1) I selected a message sent to four email addresses, more than can be listed on one line. I can see three of them. To: ends with "1 more". 

2) Thunderbird had been running full screen. I reduced its width so that only part of the first email address is listed. To: still ends with "1 more".

3) I selected a different message, and then came back to the message I'm testing. Only one email address is listed. To: ends with "3 more"

4) Changed back to full screen. Only one email address is listed. To: still ends with "3 more"

5) I selected a different message, and then came back to the message I'm testing. I can see three email addresses (just like I originally could). To: ends with "1 more". The Litmus test claims "Now you should see part of the address and "x-1 more".

I think Thunderbird is working correctly (when I reselected the message in the last step it refreshed the To: line, displaying it like it originally was) but that test seems carefully tailored to test several bugs were fixed. Please resolve the issue.

Comment 1

[rsx11m]

The display doesn't adjust when resizing the image, that's expected and possibly pending as a bug report already. I think that you are referring to bug 567062 which added a 30px threshold when the last address is to be cut off. There was a follow-up on that fix, bug 601206, which may have invalidated the original test. It's too long ago for me to tell what the issue with the first patch was to require the follow-up bug, but those bugs should hold the clue whether the litmus test has to be changed or removed altogether.

Comment 2

[rsx11m]

Ok, so bug 601206 covers a special case of bug 567062 for the "1 more" exception. Thus, https://litmus.mozilla.org/show_test.cgi?id=12638 would be applicable for changing the views between "3 more" and "2 more", but not between "1 more" and "nothing more" which is https://litmus.mozilla.org/show_test.cgi?id=13686 instead.

Another observation from Eric's description is that he was changing the views between "3 more" and "1 more" whereas either test is supposed to test the fringe cases between some "x+1" and "x" (thus only one address difference). Test #12638 states "30px to the left/right" but isn't explicit on "1 address", maybe that's the cause for not being sure what exactly to do? Test #13686 has a statement "vary the window size to be just a bit larger or smaller than needed" which could also be a more specific to "show all addresses in a single line".

Comment 3

[Thomas D. (:thomas8)]

I'm not sure what should be done here, and this bug should be in another product/component?

Comment 4

[Wayne Mery (:wsmwk)]

(In reply to Thomas D. from comment #3)
> I'm not sure what should be done here,

I'm not sure either

Comment 5

[Ludovic Hirlimann [:Usul]]

Either upd

Comment 6

[Ludovic Hirlimann [:Usul]]

ate the test or remove it from the suite

Comment 7

[rsx11m]

As said, it's a fringe case. If it's too confusing, we can remove the tests. The main reason I had added them was that all (ok, most, or many at least) patches require tests these days per policy, and I had no clue how to automate those, thus came up with plain-text descriptions for litmus testing.

I gave some hints how to improve the current wording in comment #2, with a bit speculating what might have been the reason for Eric to file this bug report.