emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [O] org-mime-htmlize: visual representation (thunderbird)


From: Eric Schulte
Subject: Re: [O] org-mime-htmlize: visual representation (thunderbird)
Date: Wed, 11 Apr 2012 09:38:46 -0400
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.1.50 (gnu/linux)

Hi Uwe,

Uwe Brauer <address@hidden> writes:

>>> On Wed, 11 Apr 2012 00:23:54 -0400, Eric Schulte <address@hidden> wrote:
>
>    > Uwe Brauer <address@hidden> writes:
>    >> 
>    >> Uwe 
>    >> 
>
>    > Hi Uwe,
>
>    > Thanks for sending along this helpful review.  I've just pushed two
>    > changes to org-mime so that it now (1) wraps html and images in a
>    > multipart/related mime structure and (2) marks images as "disposition
>    > inline" so that they don't show up as attachments.
>
> Hi Eric, 
>
> Thanks for your efforts. I have good and bad news. The bad
> news is your changes make things worse in Thunderbird, for
> reasons I don't understand the header of the resulting
> messages reads:
> Content-type: text/plain; charset=us-ascii
> which is wrong and now png are displayed!
>

OK, for my own edification I had changed the message body from
(I'm hoping these are sufficiently quoted to survive mailing)

,----[original]
| 
| html alternative... |
|   images for html...
`----

to

,----[revised (and more broken in TB)]
| 
| 
`----

which wraps the html and images into a multipart/related type.

Why is this later structure illegal?  Are nested multi type parts not
allowed?  Also, it seems that everything I've tried works in gnus and in
most web user agents.  Is thunderbird simply a stickler for the letter
of the RFC law?

>
> Which brings me to the good news. After I wrote to you I
> received a message from  the TB developers  which emphasised
> that, besides the information I have gave you, the main
> point is the header, which should be 
>
>   Content-type: multipart/related; boundary="=-=-="
>  and the thunderbird developers insist that this is the 
>  RFC 2387 standard.
>
> Gnus actually generate  via the mml-generate-mime function
> the header 
>   Content-type: multipart/mixed; boundary="=-=-="
> which is wrong.
>

OK, I've just reverted my change, but I'm keeping the change of image
disposition to "inline".

>
> I brought up the issue in the gnus mailing list and the
> developers agreed that in the case of a html message with
> png the Content-type should follow the RFC standard.
>
> I checked this explicitly: your old code but with a different
> mml-generate-mime function generates a message which is
> correctly displayed in thunderbird and GMail and Ipod for
> that manner. 
>

OK.  Then I will assume that this issue is out of my hands and wait for
gnus to change its mime wrapping behavior upstream.

>
> BTW I don't know how this issue, of the Content-type in the
> header,  is treated in VM or Wanderlust.
>

I do my best to provide reasonable implementations for these other two
mailers and assume that if anything goes wrong then someone will submit
a bug report.

>
> Now the question is how to proceed:
> I had the idea of introducing a new variable mml-mime-use-related and wrap it
> into the mml-generate-mime code. Then org-mime-htmlize
> should set this variable to t, and later a different
> function should be added to the mail-send-hook setting the
> variable to nil again.
>
> Lars didn't like the idea and came up with a different
> implementation. However I don't see how to use it easily. So
> I include both solutions and let you decide which fits best
> for org-mime-htmlize.
> But as it is now you should undo your recent changes because
> even with  the *new* mml-generate-mime function and your
> *new* code the resulting mail is not displayed correctly in
> TB.
>
> I have now added lars and the ding mailing list to the CC.
>

While Lars' solution does look cleaner it is not clear to me how this
would be used from an external tool (such as org-mime) which does not
call `mml-generate-mime' explicitly, but rather relies on the normal
mailing process to handle these specifics.  Wouldn't it make the most
sense for the mailing process to inspect the email and set the
appropriate multipart type automatically.

Thanks,

>
> Regards
>
> Uwe 
>
> My solution:
> ,----
> | (defvar mml-mime-use-related t
> | "*Variable to control whether to use `multipart/mixed' or 
> `multipart/related'.")
> | 
> | (defun mml-generate-mime ()
> |   "Generate a MIME message based on the current MML document."
> |   (let ((cont (mml-parse))
> |             (mml-multipart-number mml-multipart-number))
> |     (if (not cont)
> |             nil
> |       (mm-with-multibyte-buffer
> |             (if (and (consp (car cont))
> |                              (= (length cont) 1))
> |                     (mml-generate-mime-1 (car cont))
> |               (if mml-mime-use-related
> |                       (mml-generate-mime-1 (nconc (list 'multipart '(type . 
> "related"))
> |                                                                             
>   cont))
> |                     (mml-generate-mime-1 (nconc (list 'multipart '(type . 
> "mixed"))
> |                                                                             
> cont)))
> |     (buffer-string))))))
> `----
>
>
> Lars solution 
>
> ,----
> | (defun mml-generate-mime (&optional multipart-type)
> |   "Generate a MIME message based on the current MML document.
> | MULTIPART-TYPE defaults to \"mixed\", but can also
> | be \"related\" or \"alternate\"."
> |   (let ((cont (mml-parse))
> |     (mml-multipart-number mml-multipart-number)
> |     (options message-options))
> |     (if (not cont)
> |     nil
> |       (prog1
> |       (mm-with-multibyte-buffer
> |         (setq message-options options)
> |         (if (and (consp (car cont))
> |                  (= (length cont) 1))
> |             (mml-generate-mime-1 (car cont))
> |           (mml-generate-mime-1
> |            (nconc (list 'multipart (cons 'type (or multipart-type "mixed")))
> |                   cont)))
> |         (setq options message-options)
> |         (buffer-string))
> |     (setq message-options options)))))
> `----
>
>

-- 
Eric Schulte
http://cs.unm.edu/~eschulte/

reply via email to

[Prev in Thread] Current Thread [Next in Thread]