Allow people to circumvent W3 for HTML

Allow people to circumvent W3 for HTML

[Kai Großjohann]

I think that many people want to use something like Netscape for
viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
simple variable which people can set to achieve this.

This can be implemented by changing mailcap-mime-data such that it
checks the variable, not only looks whether the w3-foo function is
bound.

Opinions?

kai
-- 
GNU/Linux provides a nice `poweroff' command, but where is `poweron'?

Re: Allow people to circumvent W3 for HTML

[Oystein Viggen]

* [Kai Großjohann] 

> I think that many people want to use something like Netscape for
> viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
> simple variable which people can set to achieve this.

Do you mean like this?
(setq browse-url-browser-function 'browse-url-netscape)

Oystein
-- 
This message was generated by a flock of happy penguins.

Re: Allow people to circumvent W3 for HTML

[Frank Schmitt]

Oystein Viggen <oysteivi@tihlde.org> writes:
 
>>I think that many people want to use something like Netscape for
>>viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
>>simple variable which people can set to achieve this.
>
>Do you mean like this?
>(setq browse-url-browser-function 'browse-url-netscape)

I'm quite sure he meant viewing html-emails and such stuff and the above
only works for visiting links.

-- 
One Ring to rule them all, One Ring to find them,
One Ring to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.
19. Dezember 2001

Re: Allow people to circumvent W3 for HTML

[David S. Goldberg]

How I deal with text/html email:

(setq mm-discouraged-alternatives (list "multipart/related"
					"text/html" "text/richtext"))

(setq mm-automatic-external-display '("text/html"))

(setq mm-inline-override-types '("text/html"))

This results in all text/html showing up as a button.  I then do a
view-interactive if I actually want to read it.  I specify a shell
script that calls netscape with the -remote option to pop up a new
window as the viewer.  I purposely do not pass the part to netscape
automatically.  99% of the text/html I get I don't actually want to
see.  Unfortunately a lot of people at work send messages that contain
both text/plain and text/html but send it as multipart/mixed instead
of multipart/alternative so the text/html is just a redundant
attachment.

-- 
Dave Goldberg
david.goldberg6@verizon.net

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

Oystein Viggen <oysteivi@tihlde.org> writes:

> * [Kai Großjohann] 
> 
>> I think that many people want to use something like Netscape for
>> viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
>> simple variable which people can set to achieve this.
> 
> Do you mean like this?
> (setq browse-url-browser-function 'browse-url-netscape)

AFAIK, this does nothing for multipart/mixed with text/html parts?

[time passes]

I now _know_ it does nothing, for I have that setting, and Gnus still
uses W3 for displaying HTML.

kai
-- 
GNU/Linux provides a nice `poweroff' command, but where is `poweron'?

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

david.goldberg6@verizon.net (David S. Goldberg) writes:

> How I deal with text/html email:
> 
> (setq mm-discouraged-alternatives (list "multipart/related"
> 					"text/html" "text/richtext"))
> 
> (setq mm-automatic-external-display '("text/html"))
> 
> (setq mm-inline-override-types '("text/html"))
> 
> This results in all text/html showing up as a button.  I then do a
> view-interactive if I actually want to read it.

Well, people have already put the right command to use in ~/.mailcap,
so why not just use that?

Also, I want to make it so that hitting RET on the MIME button invokes
Netscape (or whatever was specified in ~/.mailcap) automatically, with
no added fuss.

My experience is that a lot of people want this behavior: for each
text/html part, they see a button, and clicking this button invokes
Netscape (or whatever).  Therefore, we should make it easy for people
to get this behavior.  (Well, de-installing W3 would be easy but
surely that's not the right approach.)

kai
-- 
GNU/Linux provides a nice `poweroff' command, but where is `poweron'?

Re: Allow people to circumvent W3 for HTML

[David S. Goldberg]

>> 
>> This results in all text/html showing up as a button.  I then do a
>> view-interactive if I actually want to read it.

> Well, people have already put the right command to use in ~/.mailcap,
> so why not just use that?

In my case I haven't put it in .mailcap but that's just because I was
too lazy.

But maybe I'm confused.  Are you saying that if I put my shell script
in ~/.mailcap then C-u <part> K e, or the equivalent option on the
button itself wouldn't do the right thing?  If so then something is
broken.  I would expect that .mailcap entries should take precedence
as the default action.

-- 
Dave Goldberg
david.goldberg6@verizon.net

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

david.goldberg6@verizon.net (David S. Goldberg) writes:

> But maybe I'm confused.  Are you saying that if I put my shell script
> in ~/.mailcap then C-u <part> K e, or the equivalent option on the
> button itself wouldn't do the right thing?  If so then something is
> broken.  I would expect that .mailcap entries should take precedence
> as the default action.

With the correct entry in .mailcap, C-u 42 K e should indeed do the
right thing.

But my suggestion was to include a configuration option such that (a)
buttons for text/html parts appear automatically, and (2) hitting RET
or mouse-2 on such a button invokes whatever is in .mailcap.

Implementing this is quite simple, I think, though I haven't tested
it.  Right now, there is the following entry in mailcap-mime-data:

/----
|      ("html"
|       (viewer . mm-w3-prepare-buffer)
|       (test   . (fboundp 'w3-prepare-buffer))
|       (type   . "text/html"))
\----

Now we introduce a new variable and say (test . (and foo (fboundp
'w3-prepare-buffer))) in the third line above.

What do you think?

kai
-- 
GNU/Linux provides a nice `poweroff' command, but where is `poweron'?

Re: Allow people to circumvent W3 for HTML

[Daniel Pittman]

On Fri, 19 Oct 2001, Kai Großjohann wrote:
> I think that many people want to use something like Netscape for
> viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
> simple variable which people can set to achieve this.
> 
> This can be implemented by changing mailcap-mime-data such that it
> checks the variable, not only looks whether the w3-foo function is
> bound.
> 
> Opinions?

Yes, please. Actually, allowing me to easily control what is used to
render HTML to text would be ideal; I want to pipe it through (for
example) html2text(1) rather than use W3.

        Daniel

-- 
Crying loud, you're crawling on the floor
Just a beautiful baby, You're nothing more
        -- Switchblade Symphony, _Clown_

Re: Allow people to circumvent W3 for HTML

[Simon Josefsson]

Daniel Pittman <daniel@rimspace.net> writes:

> On Fri, 19 Oct 2001, Kai Großjohann wrote:
>> I think that many people want to use something like Netscape for
>> viewing HTML, even if Emacs/W3 is installed.  IMHO there should be a
>> simple variable which people can set to achieve this.
>> 
>> This can be implemented by changing mailcap-mime-data such that it
>> checks the variable, not only looks whether the w3-foo function is
>> bound.
>> 
>> Opinions?
> 
> Yes, please. Actually, allowing me to easily control what is used to
> render HTML to text would be ideal; I want to pipe it through (for
> example) html2text(1) rather than use W3.

Doesn't

text/html: html2text %s; copiousoutput

or something in ~/.mailcap work?

Hopefully Netscape won't cater to that line.

Re: Allow people to circumvent W3 for HTML

[Per Abrahamsen]

How aboud doing it like this:

(defcustom mm-html-prepare-buffer (when (fboundp 'w3-prepare-buffer)
                                    'w3-prepare-buffer)
  "Function used for inline rendering html in the current buffer.

Set it to nil to disable inline html drawing."
  :type '(choice (const :tag "Disable" nil) function)
  :group 'mime-display)

/----
|      ("html"
|       (viewer . mm-html-prepare-buffer)
|       (test   . (fboundp 'mm-html-prepare-buffer))
|       (type   . "text/html"))
\----

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

Simon Josefsson <jas@extundo.com> writes:

> Hopefully Netscape won't cater to that line.

How does the `test' feature work?  I'm sure a test could be included
that prevents Netscape from looking at the line.

kai
-- 
Lisp is kinda like tpircstsoP

Re: Allow people to circumvent W3 for HTML

[Simon Josefsson]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> Simon Josefsson <jas@extundo.com> writes:
> 
>> Hopefully Netscape won't cater to that line.
> 
> How does the `test' feature work?  I'm sure a test could be included
> that prevents Netscape from looking at the line.

It is described in RFC 1524, basicly it is a external program that is
able to give a 0/1 answer.  But neither Mozilla nor Netscape seems to
listen to a text/html line in .mailcap anyway.

Re: Allow people to circumvent W3 for HTML

[Daniel Pittman]

On Sat, 20 Oct 2001, Simon Josefsson wrote:
> Daniel Pittman <daniel@rimspace.net> writes:

[...]

>> Yes, please. Actually, allowing me to easily control what is used to
>> render HTML to text would be ideal; I want to pipe it through (for
>> example) html2text(1) rather than use W3.
> 
> Doesn't
> 
> text/html: html2text %s; copiousoutput
> 
> or something in ~/.mailcap work?

Er, didn't last time I tried... ...time passes...

...or maybe not. I thought I did have it in there. Let me try again.

No. Adding it and restarting XEmacs (and, thus, Gnus) completely failed
to cause to it take effect: the HTML mail still caused W3 to load.

Lesse... Within `mailcap-mime-data', we have my values *first* in the
list, with the W3 values last.

W3 is definitely chosen; there isn't an issue with searching this data
in reverse order, is there? More testing...

Found it. The breakage comes for `mm-display-inline'

Basically, `mm-display-part' comes in, makes sure the mailcap files are
parsed, locates the viewer from the mailcap data (getting html2text),
then asks:

  (if (and (mm-inlinable-p handle)
           (mm-inlined-p handle))

When this passes, it calls `mm-display-inline' which unconditionally
uses W3. The HTML part isn't considered to be capable of being inlined
unless W3 is present.


Now, having looked at this, I can most certainly go and fix this problem
by redefining the stuff in the inlined type data structures to call my
shell command and parse the content that way.

However, in answer to your original question: No, adding that to
.mailcap or *any* other mailcap file will never work when the W3 library
is installed.

So, bug or feature? You decide ... but I would suggest adding a fairly
easy to reach default that let the end user call a shell command to
reformat the text and make it inline rather than forcing W3.


I will post my solution once it's written; it should be capable of doing
that if someone with commit access wants to add it. It waits on my
finishing cleaning my bedroom so I have somewhere to sleep tonight,
though. :)


> Hopefully Netscape won't cater to that line.

It seems to be brighter than that, at least for versions of Netscape
equal to Mozilla 0.9.5+. :)

      Daniel

-- 
Time spent in the advertising business seems to create a
permanent deformity like the Chinese habit of foot-bonding.
        -- Dean Acheson

Re: Allow people to circumvent W3 for HTML

[Simon Josefsson]

Daniel Pittman <daniel@rimspace.net> writes:

>> Doesn't
>> 
>> text/html: html2text %s; copiousoutput
>> 
>> or something in ~/.mailcap work?
> 
> Er, didn't last time I tried... ...time passes...
> 
> ...or maybe not. I thought I did have it in there. Let me try again.
> 
> No. Adding it and restarting XEmacs (and, thus, Gnus) completely failed
> to cause to it take effect: the HTML mail still caused W3 to load.
> 
> Lesse... Within `mailcap-mime-data', we have my values *first* in the
> list, with the W3 values last.
> 
> W3 is definitely chosen; there isn't an issue with searching this data
> in reverse order, is there? More testing...
> 
> Found it. The breakage comes for `mm-display-inline'
> 
> Basically, `mm-display-part' comes in, makes sure the mailcap files are
> parsed, locates the viewer from the mailcap data (getting html2text),
> then asks:
> 
>   (if (and (mm-inlinable-p handle)
>            (mm-inlined-p handle))
> 
> When this passes, it calls `mm-display-inline' which unconditionally
> uses W3. The HTML part isn't considered to be capable of being inlined
> unless W3 is present.
> 
> 
> Now, having looked at this, I can most certainly go and fix this problem
> by redefining the stuff in the inlined type data structures to call my
> shell command and parse the content that way.
> 
> However, in answer to your original question: No, adding that to
> .mailcap or *any* other mailcap file will never work when the W3 library
> is installed.
> 
> So, bug or feature? You decide ... but I would suggest adding a fairly
> easy to reach default that let the end user call a shell command to
> reformat the text and make it inline rather than forcing W3.

I'd say bug, if you say something in .mailcap, Gnus should use that
unless you've told it otherwise somewhere else or if there is a good
reason to do something else.

> I will post my solution once it's written; it should be capable of doing
> that if someone with commit access wants to add it.

IMHO even if nothing is specified in .mailcap, Gnus should use a
customizable variable to say whether w3, w3m or some other Emacs web
browsers should be used instead of hardcoding it to w3.  So you should
be able to override the text/html viewer used either by .mailcap or by
changing the variable (the latter having more priority).

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

Daniel Pittman <daniel@rimspace.net> writes:

> When this passes, it calls `mm-display-inline' which unconditionally
> uses W3. The HTML part isn't considered to be capable of being inlined
> unless W3 is present.

But there are variables that you can use to tell Gnus that text/html
is not inlineable.  (Inlinable?)

kai
-- 
Lisp is kinda like tpircstsoP

Re: Allow people to circumvent W3 for HTML

[Henrik Enberg]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> How aboud doing it like this:
> 
> (defcustom mm-html-prepare-buffer (when (fboundp 'w3-prepare-buffer)
>                                     'w3-prepare-buffer)
>   "Function used for inline rendering html in the current buffer.
> 
> Set it to nil to disable inline html drawing."
>   :type '(choice (const :tag "Disable" nil) function)
>   :group 'mime-display)

This, I think is the way to go.  Then I could write my owm function to
call w3m.el instead.  I currently have to have w3 installed just so I
can defadvice `mm-inline-text' to use w3m.el.

Henrik
-- 
I'm CONTROLLED by the CIA!!  EVERYONE is controlled by the CIA!!

Re: Allow people to circumvent W3 for HTML

[Daniel Pittman]

On Sat, 20 Oct 2001, Kai Großjohann wrote:
> Daniel Pittman <daniel@rimspace.net> writes:
> 
>> When this passes, it calls `mm-display-inline' which unconditionally
>> uses W3. The HTML part isn't considered to be capable of being
>> inlined unless W3 is present.
> 
> But there are variables that you can use to tell Gnus that text/html
> is not inlineable.  (Inlinable?)

Er, yes... but won't that stop me doing what I want?

I *want* HTML parts to be inline. I have enough people who can only send
HTML mail without plain text parts[1] that I can't just disable it the
way I want. The price of having end users, don't you know. ;)

So, I want HTML rendering and I want it to be done inline. Sending it to
an external process that required going to another window would be
highly filled with suckage.

So, no, I don't think "not inline" is the right way to set up my Gnus
rendering, is it?

        Daniel

Footnotes: 
[1]  Thank you *so* much, Microsoft.

-- 
Of course I lie to people. But I lie altruistically - for our mutual
good. The lie is the basic building block of good manners. That may
seem mildly shocking to a moralist - but then what isn't?
        -- Quentin Crisp

Re: Allow people to circumvent W3 for HTML

[Per Abrahamsen]

Henrik Enberg <henrik@enberg.org> writes:

> This, I think is the way to go.  Then I could write my owm function to
> call w3m.el instead.

And submit it as a patch to be included in Gnus ;-)

Re: Allow people to circumvent W3 for HTML

[Kai Großjohann]

Daniel Pittman <daniel@rimspace.net> writes:

> On Sat, 20 Oct 2001, Kai Großjohann wrote:
>> Daniel Pittman <daniel@rimspace.net> writes:
>> 
>>> When this passes, it calls `mm-display-inline' which unconditionally
>>> uses W3. The HTML part isn't considered to be capable of being
>>> inlined unless W3 is present.
>> 
>> But there are variables that you can use to tell Gnus that text/html
>> is not inlineable.  (Inlinable?)
> 
> Er, yes... but won't that stop me doing what I want?

Oops, of course you're right.  Sorry.

[time passes]

But can't you augment mm-inline-media-tests with something that calls
w3m.el?  Maybe replace the text/html entry rather than augmenting it.

kai
-- 
Lisp is kinda like tpircstsoP

Re: Allow people to circumvent W3 for HTML

[Oystein Viggen]

* [Kai Großjohann] 

> AFAIK, this does nothing for multipart/mixed with text/html parts?

Ah, sorry.  I thought you meant when clicking on buttonized URLs.

Personally, I never render html in mail or news messages, and I'm quite
happy with it that way.  But I figure that was not what you wanted,
either  :)

Oystein
-- 
When in doubt: Recompile.

Re: Allow people to circumvent W3 for HTML

[Lloyd Zusman]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> Henrik Enberg <henrik@enberg.org> writes:
>
>> This, I think is the way to go.  Then I could write my owm function to
>> call w3m.el instead.
>
> And submit it as a patch to be included in Gnus ;-)

If you do that, you could take a look at the functions I wrote and
posted a couple days ago that do almost the same thing you're talking
about.  Look under this article here in the ding@gnus.org mailing list
(the earlier articles in that thread discuss exactly what I'm trying
to do with w3m rendering of html message parts):

 From: Lloyd Zusman <ljz@asfast.com>
 Message-ID: <m2g08fdw8y.fsf@asfast.com>
 Subject: Re: How to do something a bit complicated with inline decoding?

In addition to these functions, I have this in my .gnus.el.

 ;; Make sure that Gnus never tries to render HTML or
 ;; rich text.
 (setq mm-discouraged-alternatives
       '("text/html"
 	 "text/richtext"))
 (setq mm-inline-override-types mm-discouraged-alternatives)
 (setq mm-automatic-display (remove "text/html" mm-automatic-display))


Also, in the same thread, you could check this message, as well,
for another, similar function.

 From: Raymond Scholz <ray-2001@zonix.de>
 Message-ID: <lgd.87669amwg0.fsf@mde1.zonix.de>


Contained within all these functions is almost everything that's
needed for this patch that's being discussed.  I think that all
that would have to be added is the connection to the mime types
in mailcap, and perhaps a `defcustom' entry.


-- 
 Lloyd Zusman
 ljz@asfast.com