Problems with inlining images in Emacs 20

Problems with inlining images in Emacs 20

[Barry Fishman]

I'm not an expert at gnus mime processing but:

Concerning mm-decode.el:

It seems wrong to put the image file wildcard "image/.*" as a part of
mm-inlined-types.  This seems to mean that any image type not caught
by the mm-inline-media-tests will be just placed inline.  Shouldn't
one assume that if an image is not of a known type, the Emacs 21's
will probably not know what to do with it, and if they did, at least
the mm-valid-and-fit-image-p test should be made.

In Emacs 20, inlined images do not look very pretty, and since they
are inlined there isn't even a button to save them to a file or even
get rid of them.  This is a real pain.


Concerning gnus-art.el:

In gnus-display-mime, The mm-dissect-buffer test is always done in its
strict mode.  There seems to be messages that get correctly dissected
with the strict option off, but using

     (or (mm-dissect-buffer) (mm-uu-dissect) (mm-dissect-buffer t))

about line 3776 seems to cause emacs to break (although not to crash).
Is this why strict mime processing is not a customization option?

I have been working on a `gnus-force-mime' which can be called
interactively, when you are seeing a message which is obviously
intended to be mime, but is missing something like a mime version
header.  It would be available like "C-u g" which in a sense does the
opposite in shutting off all processing.  Does this seem reasonable?
Is something like this already done?

In general:

I don't like encouraging the use of invalidly formatted mime messages,
by just having mail/news readers accept them without comment.
However, I can't just ignore mail that is incorrectly formatted, and
disassembling them by hand is a real pain.

Barry Fishman

Re: Problems with inlining images in Emacs 20

[ShengHuo ZHU]

Barry Fishman <barry_fishman@acm.org> writes:

> I'm not an expert at gnus mime processing but:
> 
> Concerning mm-decode.el:
> 
> It seems wrong to put the image file wildcard "image/.*" as a part of
> mm-inlined-types.  This seems to mean that any image type not caught
> by the mm-inline-media-tests will be just placed inline.  Shouldn't
> one assume that if an image is not of a known type, the Emacs 21's
> will probably not know what to do with it, and if they did, at least
> the mm-valid-and-fit-image-p test should be made.

Emm. I think the recent change of "Default to displaying as text"
causes the problem.  I've fixed it.

> In Emacs 20, inlined images do not look very pretty, and since they
> are inlined there isn't even a button to save them to a file or even
> get rid of them.  This is a real pain.

> Concerning gnus-art.el:
> 
> In gnus-display-mime, The mm-dissect-buffer test is always done in its
> strict mode.  There seems to be messages that get correctly dissected
> with the strict option off, but using
> 
>      (or (mm-dissect-buffer) (mm-uu-dissect) (mm-dissect-buffer t))
> 
> about line 3776 seems to cause emacs to break (although not to crash).
> Is this why strict mime processing is not a customization option?

I don't think so. Forcing to dissect a non-strict mime message may
causes some unexpected troubles.

> I have been working on a `gnus-force-mime' which can be called
> interactively, when you are seeing a message which is obviously
> intended to be mime, but is missing something like a mime version
> header.  It would be available like "C-u g" which in a sense does the
> opposite in shutting off all processing.  Does this seem reasonable?
> Is something like this already done?

`K m' repairs it.

> In general:
> 
> I don't like encouraging the use of invalidly formatted mime messages,
> by just having mail/news readers accept them without comment.
> However, I can't just ignore mail that is incorrectly formatted, and
> disassembling them by hand is a real pain.

`K m', `W 6', `W q' and `W h' are designed for this purpose.

ShengHuo

Re: Problems with inlining images in Emacs 20

[Barry Fishman]

ShengHuo ZHU <zsh@cs.rochester.edu> writes:
>
> Barry Fishman <barry_fishman@acm.org> writes:
>> Concerning mm-decode.el:

> Emm. I think the recent change of "Default to displaying as text"
> causes the problem.  I've fixed it.

Now all my problem messages work fine, unexpectedly without a 'K m'.
Thanks.

>> In general:
>> 
>> I don't like encouraging the use of invalidly formatted mime messages,
>> by just having mail/news readers accept them without comment.
>> ...

> `K m', `W 6', `W q' and `W h' are designed for this purpose.

As usually your n steps in front of me.  Everything was in the
documentation (and reference card).  (So I can add 'K b').  Can I
complain that there are so many feature that I don't expect, I just
forget to look for them? :-)  I need to read through the manual again.
It seems to take continual reading since many parts don't make sense
until you start using the feature.

Thanks Again
Barry