[imp] Signature

azurIt azurit at pobox.sk
Tue Jan 22 10:33:34 UTC 2013


> Od: "Michael M Slusarz" <slusarz at horde.org>
> Komu: <imp at lists.horde.org>
> Dátum: 21.01.2013 21:42
> Predmet: Re: [imp] Signature
>
>Quoting Ralf Lang <lang at b1-systems.de>:
>
>> I understand the arguments for and against showing/editing signatures.
>> Still I do not find it very intuitive and when H5 goes into production
>> this month, I am sure users will ask for patches to at least see the
>> signature.
>
>To summarize dozens of previous posts why signatures can NOT be added  
>to compose text:
>
>1.) A signature MUST MUST MUST be added to the bottom.  For those that  
>top-post on replies, your signature MUST NOT appear after "your" text,  
>because then there is no way of telling below the signature what is  
>YOUR data vs. what is OTHER'S data.  You can't allow users to change  
>this, period.
>
>If someone tells me this is how outlook/Gmail/some other MUA does it,  
>I will slap them.
>
>2.) Signatures have an "unofficial" delimiter that has a specific  
>format (although unofficial, it is reference in at least one official  
>RFC).  A user MUST NOT be able to change this delimiter.   
>Additionally, this delimiter can get munged during draft  
>saving/resuming, or browser refreshes/sending.  This kind of  
>structural element should never be exposed to the user.
>
>3.) Resuming drafts.  It is possible to keep track if IMP adds a  
>signature to a draft message, but other MUAs have no knowledge of  
>this.  By not adding a signature to a message, it at least fixes the  
>issue of a draft message saved by IMP and resumed by another MUA from  
>having duplicate signatures.
>
>4.) Regardless of #3... a signature should NEVER EVER EVER be added to  
>a draft message since a signature CAN BE CHANGED BETWEEN DRAFT SAVE  
>TIME AND SEND TIME.  A signature is potentially dynamic.  A USER  
>SHOULD NEVER BE EXPECTED TO MANUALLY CHANGE A SIGNATURE IN THIS  
>MATTER.  "Placeholder template code/text" is not an option (see #3).
>
>5.) SIGNATURES ARE NOT SUPPOSED TO BE EDITED PER MESSAGE!!  This is  
>NOT what a signature is.  This is a fundamental misunderstanding of  
>what a signature is supposed to do (and the point most other MUAs  
>miss).  This describes instead a "template" or "auto-append" feature  
>rather than a signature.
>
>People keep bringing up the idea that "users want to alter their  
>signature".  Why?  Someone give me ONE REAL-WORLD EXAMPLE OF WHEN A  
>USER WOULD CONCEIVABLY DO THIS.
>
>Here's an example signature:
>
>John Smith
>Manager, Widgets LLC
>Email: foo at exaple.com
>Mailing Address: 123 Main Street
>                  Anytown, Anystate 12345 USA
>
>What part of that signature would a *realistic* user change on any  
>given e-mail message?
>
>The only thing I could potentially see is sending a message using a  
>"personal" signature vs. a "business" signature, and needing different  
>signatures for both.  But that is the entire point of multiple  
>identities in the first place.
>
>6.) Signatures make no sense to add in the minimal or smartmobile  
>view.  In both of these cases, signature information (which may be  
>quite long) simply detracts from being able to actually enter text to  
>send a message.  So now you have the case where you have inconsistent  
>UI between different views (some show signature, some don't).  That is  
>*VERY* confusing for users.  If UI elements are shared between views,  
>they must act similarly.  The only practical answer is to never show  
>the signature data.
>
>7.) It is IMPOSSIBLE to reliably switch signature data when switching  
>between HTML/Plaintext, and switching between identities.  The code in  
>previous versions of IMP DID NOT EVEN PRETEND THAT THIS WOULD RELIABLY  
>HAPPEN.  There's nothing worse than writing code that "might" work  
>depending on how the user has altered the compose text.  The amount of  
>time spent trying to maintain this terrible mess of code far exceeded  
>the benefit it ever provided.
>
>
>That's *7* reasons why this is a terrible idea.  I could go on more -  
>such as with how the real-world example of letterhead, quoted as  
>reason why signatures need to be displayed, actually is a slam-dunk  
>argument as to why signatures should not be displayed - but I have  
>better things to do today rather than continuing this argument that is  
>well-settled.
>
>This is the last time I will discuss this topic.  Rest assured, it is  
>not happening in IMP ever so everyone can stop wasting their time  
>bringing it up.  This time is better spent patching IMP locally as you  
>see fit (the benefit of open-source).



After reading this, it looks to me that you EXACTLY know what signature is, how it should looks and how it must be used by everyone. I assume you are basing your arguments on RFC or something like that. Can i see it? Thank you.

azur


More information about the imp mailing list