Changing identity with new signature results in separator and signature text assuming last paragraph format of message body

NEW
Unassigned

Status

MailNews Core
Composition
6 years ago
2 years ago

People

(Reporter: Lewis Rosenthal, Unassigned)

Tracking

({regression})

Trunk
x86
All
regression

Firefox Tracking Flags

(Not tracked)

Details

(Whiteboard: dupme)

(Reporter)

Description

6 years ago
Replying to a message with quoting enabled and the signature set to display below the quote, the default identity places signature in the correct place. However, changing identities results in the new signature being placed within the quoted text (at the bottom).

Reproducible: Always

Steps to reproduce:

1. Enable HTML format for composition.
2. Enable Automatically quote & start reply above quote.
3. Enable Place signature below the quote.
4. Enable Include signature for replies.
5. Ensure that there is at least one additional identity configured with a signature, and that the default identity has a signature.
6. Reply to a message.
7. Change to the/an alternate identity.
8. Scroll to bottom of message to view signature within quoted text.

Expected result: New signature should replace default signature in the same location, below quoted text.

Actual result: New signature appears above correct location, appended to quoted text.
(Reporter)

Comment 1

6 years ago
Build identifier: Mozilla/5.0 (OS/2; Warp 4.5; rv:8.0) Gecko/20111023 Firefox/8.0 SeaMonkey/2.5

Specifics: 2.5b3

Updated

6 years ago
Hardware: Other → x86

Comment 2

6 years ago
Lewis, are you able to test this on another platform?  I suspect it's not an OS/2 specific issue.
(Reporter)

Comment 3

6 years ago
Confirmed that it does *not* occur under Linux (openSUSE 11.4) under either 2.5b3 or 2.5b4 as available from official Mozilla.org ftp (i.e., not my own build).

To verify the occurrence of this under OS/2, I used the default theme, and ran under safe mode (no extensions whatsoever). The issue presents itself whether or not the signature is in a file to be included or specified in the identity itself. It also is consistent in this behavior whether the signature is specified as HTML or not, and whether or not a previous signature is present in the message to which the reply is being made (i.e., it appears that "-- " in the original message is *not* a factor).

I have not tested under Win32 as yet, but as I'm not seeing this on Linux, I would suspect that we are the only ones seeing it.

Possibly related: When replying within a quoted block, <Enter> does not feed an extra line, and instead, cuts the quoted text off at the end of the given line. As an example, consider the following:

| This is some quoted text, followed by a paragraph break.
|
| This is a new quoted paragraph.

Replying to the above, and pressing <Enter> on the line *between* the paragraphs, yields the following:

| This is some quoted text, followed by a paragraph break.
My new text.

|
| This is a new quoted paragraph.

vs the previous (2.0.x and prior; not sure of 2.1 or 2.3) behavior:

| This is some quoted text, followed by a paragraph break.
|
My new text.

| This is a new quoted paragraph.

I think this is related (i.e., lack of newline following quoted text) to the signature symptom. Can anyone confirm for me please? I might be able to fix this in a user css setting as a temporary workaround or propose a permanent fix.

Thanks, guys!
(Reporter)

Comment 4

6 years ago
The quoting issue I cited above *only* seems to apply to plain text (original) messages. The signature issue occurs in either case. Also, neither issue cares whether the new (reply) message is composed in HTML format or plain text.

In addition, the following prefs seem to have no effect on the signature issue:

mail.quoteasblock
mail.quoted_graphical

and adding:

blockquote[type=cite] {
    padding-bottom: 1 !important;
    }

to userContent.css has no impact, either.

That said, adding *anything* on the blank line *above* the signature (another line break or even a single character) will cause the switch in signatures to occur correctly (i.e., new signature replacing old one, at the bottom of the message, below the quoted text - and the "new" text). Thus, I don;t think anyone replying to a message with the setting to begin typing below the quoted text will see this issue.

Geez, this stuff is hard to describe! ;-)
(Reporter)

Comment 5

6 years ago
I am still seeing this on:

Mozilla/5.0 (OS/2; Warp 4.5; rv:8.0.1) Gecko/20111123 Firefox/8.0.1 SeaMonkey/2.5

(2.5 GA)

Comment 6

6 years ago
Do you also see this in Thunderbird 8?
(Reporter)

Comment 7

6 years ago
[Changing this to MailNews Core]

Confirmed, Philip; I just installed TB8, which has a configuration for my IMAP account. I added a signature to my default identity (in the pref box, not via an attached file), and then added a second identity with a unique signature. Neither signature entry had any whitespace above it or to the left (i.e., no padding).

I opened a message in my Inbox, clicked reply, and noted that my default signature was positioned as it should have been, below the quoted text, preceded by a blank (unquoted) line above the signature separator. The signature was recognized by TB as a signature (i.e., lower contrast).

I then changed the identity by clicking in the From header box, exposing the dropdown list, and clicking on the alternate identity. I scrolled to the bottom of the message to find my new signature within the quoted text (i.e., below the original quoted text, but without a blank line above the signature separator, so the new sig appeared to be part of the previous - quoted - message, and styled as quoted text).

So, this does not appear to be specific to SM.

However, the accompanying condition I referenced in comment #3 does not appear to happen under TB, so the two may indeed be separate issues.

Thanks for following up.
Component: MailNews: Composition → Composition
Product: SeaMonkey → MailNews Core
Version: SeaMonkey 2.5 Branch → Trunk
(Reporter)

Comment 8

6 years ago
With all of the crazy versioning these days, I'm not sure if I set the version right (Trunk?) in this bug for SM 2.5 / TB 8.0. Please correct if I missed the mark.

Thanks, and apologies for the bugspam!
Whiteboard: dupme

Comment 9

6 years ago
I'm not sure if I'm reporting on the correct bug but I do see the exact behaviour Lewis describes on TB 8.0 on windows 7 64bit.
(Reporter)

Comment 10

6 years ago
Actually, this bug needs to be renamed/expanded.

I just created a new email, which included some indented text. To do that, I used the Ctrl+] (Format | Increase Indent) formatting command. My last line of text was still in the indented text, followed by a single blank line and my default signature, flush left.

Changing my identity switched my signature, and resulted in indenting the separator (-- ) and the signature text beneath.

As a further test, I employed a couple other formatting options. Character formatting is not applied to the separator or following text, but paragraph formatting is. Centering text, for example, and then changing identities with a different signature, will yield the same paragraph formatting applied to the signature text as the last line of text in the body of the message.

Thus, it appears that this behavior is consistent for paragraph formatting and not just for quoted text.

Build identifier: Mozilla/5.0 (OS/2; Warp 4.5; rv:9.0.1) Gecko/20111221 Firefox/9.0.1 SeaMonkey/2.6
(Reporter)

Updated

6 years ago
Summary: Changing identity with new signature when replying moves signature into quoted text → Changing identity with new signature results in separator and signature text assuming last paragraph format of message body
(Reporter)

Comment 11

6 years ago
I just tested under TB 3.0.11 running on W2K. In that instance, TB *deliberately* inserts a hard break when changing the identity, and the formatting stays intact.

Testing again on TB 9.0 on openSUSE (as well as SM 2.6.1 and 2.7 and TB 10.0 on openSUSE), I see the same behavior as SM 2.6 on OS/2, where the new sig inherits the formatting of the preceding paragraph.

Considering this, and Karl's observation in comment 9, I am changing this from OS/2 to all platforms (if someone can test on a Mac, I will gladly adjust).
OS: OS/2 → All

Comment 12

6 years ago
I see the same issue under TB 10.0.1 on Win7.
I use "Virtual Identity" add-on that switches identities on reply, so this issue is triggered immediately on each reply and is even more annoying.
This issue is similar to bug 614919, but triggered after identity switch.
Is there any workaround?
(Reporter)

Comment 13

6 years ago
And I see the issue using Flexible Identity, which is configured to prompt before making the change. The only workaround I've found, Ivan, is to deliberately go to the line above the sig separator and insert a hard break via <Enter>. If you can configure Virtual Identity to prompt before making the identity switch, that would give you an opportunity to insert the break before this bug kicks in.

Annoying, isn't it?

FWIW, I can also confirm this on TB 10.0 under OS/2.
This is a duplicate - which I can't find now.
(Reporter)

Comment 15

6 years ago
Thanks for the suggestion, Ludovic. ;-)

Some digging turns up:

Bug 623745 - marked as duplicate of:
Bug 614919 - which is marked fixed, and also references a dupe:
Bug 619028 - and also,
Bug 615377 - tagged as a dupe of Bug 614919

This does indeed look (to me, at least) like what was described in Bug 614919.

So, for those more familiar with bug etiquette than I, what is the proper procedure to query those bug reporters to see if this is indeed a regression? Dupe this over to Bug 614919 and re-open?

Comment 16

6 years ago
Maybe related: when this bug is triggered, cursor is incorrectly placed at the end of signature. This breaks "Tab" behavior, because cursor is expected to be above quotation. This is described in bug 118050.
(Reporter)

Comment 17

6 years ago
Confirmed, Ivan, and thanks for the reference to that (wow!) 10-year-old Bug 118050. Until now, I hadn't noticed this behavior, because I usually compose first, and then change the identity.

I'd love to be able to dupe this to Bug 614919 with your reference to Bug 118050 (possibly related).
(In reply to Lewis Rosenthal from comment #15)

> So, for those more familiar with bug etiquette than I, what is the proper
> procedure to query those bug reporters to see if this is indeed a
> regression? Dupe this over to Bug 614919 and re-open?

No as it's been fixed we can then keep this one as somehtin might have regressed again.
Status: UNCONFIRMED → NEW
Ever confirmed: true

Comment 19

6 years ago
> No as it's been fixed we can then keep this one as somehtin might have regressed again.
Setting regression keyword.
Keywords: regression

Comment 20

5 years ago
test with TB 13.1 under windows and linux

the problem is linked with the identity switch while replying to a mail
with the standard identity the signature is OK

in my case (reply above citation  and signature under reply above citation)
after replying i get

* first line in normal (black) body format
* second line in signature (grey) format  sometimes with "--"
* third line with signature text

case 1)
if i switch identity for the reply before typing anything
i LOST the normal (black) first line
* the first line is now in grey signature format
* the second line is the signature

case 2)
i write something before switching identity
the old signature (that was after my reply) is erased
and the new signature is put top of the body BEFORE my reply

?? link with bug 604986 703086 118050 736551
Duplicate of this bug: 736551
Removing myslef on all the bugs I'm cced on. Please NI me if you need something on MailNews Core bugs from me.
You need to log in before you can comment on or make changes to this bug.