The implementation of bug #58406 created the hidden preference mail.identity.default.suppress_signature_separator. This preference variable should have a user interface for setting it. Bug #58406 resulted from user complaints about the "-- " signature separator. Although that RFE bug was implemented at least by Thunderbird 5, I very recently (3 days ago) saw such a complaint in the mozilla.support.thunderbird newsgroup. Without a visible user interface, such complaints might continue.
I'm wondering if such a preference should only toggle the default value. It may be more appropriate to do this on a per-identity basis in either the main page where the signature text is entered, or in the Composition & Addressing account settings. This would bring it closer to the other signature options. Leaving the default as a hidden preference would be consistent with the other identity-related preferences, where it sure is another question if those should be exposed as global preferences somewhere in the UI in general to simplify the setup of multiple accounts and identities with similar non-default options.
I purposely left unstated any specific details about implementation. The preference variable is currently global. In order to gain developer support for this RFE, I did not want to introduce any changes in that variable's current implementation. If the current implementation were to be changed, my personal preference would be to implement this on a per-account basis since I don't use identities. Thus, I would prefer to see an option -- a checkbox -- on the main Account Settings window. Others, however, might want to see per-identity over-rides of per-account settings. For now, my preference is to leave the preference global with a checkbox accessible from the Options window, perhaps from the [Composition > General] pane.
All mail.identity.* preferences are implemented for identities, where *.default.* specifies the value if no *.id#.* specific for an identity is provided. So, this is already considered in the bug 58406 implementation. The reason for suggesting to change the default is usually so that the user doesn't have to look up the identity number(s) associated with a specific account to change it when done manually in the Config Editor. > If the current implementation were to be changed, my personal preference > would be to implement this on a per-account basis since I don't use > identities. Thus, I would prefer to see an option -- a checkbox -- on the > main Account Settings window. Others, however, might want to see > per-identity over-rides of per-account settings. Without the need for changing the backend, such a checkbox could be provided in the main account page as well as in the main tab for each identity as part of the "Manage Identities" dialog, so this is certainly possible (and would also be my preference for the reasons stated in comment #1 to keep things together).
I'm not a fan of the million-checkboxes approach, perhaps we can do something a little smarter instead? :) Would someone like to take a poll in mozilla.support.thunderbird (and other relevant places), and ask what people want their signatures to be, so that we can try and figure out whether we can be a little more intelligent about adding the "-- "? (I'm hoping we can implement something like "If the signature is plain-text, and all the lines start with alphanumeric characters, then add the delimiter, otherwise don't." but I don't have enough examples to tell whether that fix most of the problems people are having…) Thanks, Blake.
(In reply to Blake Winton (:bwinton - Thunderbird UX) from comment #4) > (I'm hoping we can implement something like "If the signature is plain-text, > and all the lines start with alphanumeric characters, then add the > delimiter, otherwise don't." but I don't have enough examples to tell > whether that fix most of the problems people are having…) Why the alphanumeric rule ?
(In reply to Ludovic Hirlimann [:Usul] from comment #5) > > (I'm hoping we can implement something like "If the signature is plain-text, > > and all the lines start with alphanumeric characters, then add the > > delimiter, otherwise don't." but I don't have enough examples to tell > > whether that fix most of the problems people are having…) > Why the alphanumeric rule ? Because: Blake Winton Thunderbird UX Lead 123 Fourth St. Portland, Oregon 90210 should get a separator, but: Later, Blake. -=-=-=-=-=-=-=-=- Pithy saying here! -=-=-=-=-=-=-=-=- probably shouldn't.
Re comment #4: This is definitely more complicated than what I requested when I submitted this bug report. I would want a simple Boolian decision: Does the user want "-- " or not? One problem with automating according to whether ALL lines begin alphanumerically would be explaining to users what "alphanumeric" means. I am quite sure that such an implementation would elicit never-ending questions of why (or why not) signature blocks are stripped in replies and forwards.
It may be more complicated to code, but it would be less complicated for the user. (And I would "r-" anything which presented the word "alphanumeric" to the user, since that's needless technical jargon. I would expect the documentation to say something along the lines of "If you don't want Thunderbird to add the standard signature separator, just put in one of your own, such as '-=-=-=-=-=-'.") Finally, the alphanumeric test was just an example of something that might cover the cases. Until we know what people want to end their email with, it will be hard to be intelligent about figuring out when we should add the "-- ".
This reminds me of the discussion in bug 324495 on adding the signature textbox to the account settings, whether or not an HTML input should be detected by some fancy algorithm or if an explicit "Use HTML" box should be used. The discussion there ended up with the additional checkbox, given the reasoning that "automagically" set functionality may be hard to understand for a user. Similarly, the HTML-autodetect algorithm to decide whether or not a text/html part is to be sent has been (and still is) criticized for missing cases, and a bug for switching off that automagicality is pending as bug 136502 for years. Thus, I tend to agree with David that a "keep it simple" approach should be more beneficial than creating yet another fancy algorithm which may or may not meet the user's intentions. If you nevertheless go with such, at least it should come with an off switch to allow users to control the separator manually, i.e., you'll need the checkbox in the end after all.
You need to log in before you can comment on or make changes to this bug.