Re: [Sheflug] HTML mail (was Re: Linux)

[Alex Hudson <xxxxx [at] xxxxxxxxxxxxxx>]

On Fri, Sep 20, 2002 at 04:07:46PM +0100, Will Newton wrote:
> A proper client can handle this for you. A line wrapping algorithm comes in 
> at a fair few less than an HTML parser. 

Except, you can't do it. By the nature of ascii text, the formatting is
inextricably linked to the content - change the content, and you change
the formatting. The machine cannot know whether or not the change makes
sense unless you write in block paragraphs.
- Not everyone writes in block paragraphs;
- occasionally people insert stuff other than text, such as 'ascii art'
in an attempt to work around the limitations of ascii.

> > <snip quoting>
> Again, a good client can do this for you.

Not client gets it right 100% of the time though. They're all 95% hacks,
which break down on complex e-mails - precisely those which you need 
the feature most for! Very annoying. I can re-format basic e-mails by
hand, like this one. Complex ones I can't, too time consuming, that's 
where I need the machine to get it right.

> Yeah, but... I can't say as I've ever wanted to. You can't do pie charts in 
> HTML but I don't see it being a problem.

Tables are several orders of magnitude more useful and common though.

Hi Alex, Here is your book order:

	Book			Value
	Fly fishing, JR Hartley	£39.99
	Nigella cooks dinner	£19.99
	Floyd on Fire		£24.98
		Subtotal:	£lots

What a hack. Cutting and pasting is miserably broken, so I cannot edit
that data - it's just there, in stone, forever, unless I muster the 
willpower to spend time to craft it into shape.

> > 4. Collaborative e-mail editing.
> 
> Not sure what you mean by that. Sounds more like a tool issue to me.

[Comments in red by Alex]

<p>And in closing, may I take this opportunity to tell you how bad your
service has been.<red> And you smell.</red> Good riddance to bad 
rubbish!</p><p>Yours, etc.,</p>

Etc. Highlighting, track changes. You can't do that with text.

> > <snip i18n>
> That's a different issue altogether. UTF-8 deal with this fine. Admittedly 
> UTF-8 capable clients are far and few between, but it beats the cobbled 
> together crap that is HTML entities.

You can write HTML documents in UTF-8, and I bet more browser components
would handle it correctly than mail components :P

> Anyhoo, the amount of "HTML" that I read 
> that has MS "Smart" Quotes etc. in it is unbelievable.

That doesn't count as HTML though. If there were one dominant,
proprietary text client in the same manner as there is one dominent 
HTML client, then you would suffer the same idosyncrasies.

> OK, HTML is hard to render on text consoles which a fair number of people use.

Define 'hard'. You can get away with one that understands '<p>' and
'<br>' alone (I know someone who wrote one for the BBC micro doing 
just that!), and both tags will do roughly the same thing ;)

> HTML is not designed for email, how for example may one link to a previous 
> email?

Straw man. You can't do that with text e-mail either. And while we're on
the subject of links, it's bloody text that has inflicted us with the
"Wrath[1] of[2] Footnotes[3]". We have <a> tags for this, they are very
standard and easy to implement.

> There is no nestable <quote> tag.

<blockquote> nests over here.

> It's just not designed for email. 

Neither was ASCII. Both are designed so that people can understand the
information within certain parameters; it's hard to say that either was
designed for email.

> The very nature of it makes it too complicated to compose email in so it has 
> to be done opaquely through some form of WYSIWYG editor, which again is 
> impossible on text interfaces and produces generally large and 
> incomprehensible code.

It doesn't have to be done through a WYSIWYG editor. txt2html does a
fine job (most of the time ;) at a basic level, although obviously you
would miss out on many of HTML's fine features if you were to edit in 
text mode rather than a superior native mode.

> HTML needs a parser, and a complicated one at that.

It needs a parser, granted. How complicated it needs to be depends on
the client. A text client need not interpret much more than line
breaks and links, and you can get a full text-based web browser, 
including cookies, mouse support, frames, tables, under 700k. If
you're using Emacs, it's all built in anyway :D

> is not rigourously specified (XHTML does this, but no-one uses it yet) so 

www.alexhudson.com uses X/HTML1.1. Perhaps I ought to hack in 
Evolution support :)

> people are coralled into using, so no client can ever really claim to "read 
> HTML email" apart from Outlook which dictates to everyone else.

That's an argument against proprietary monopolies, not a technical 
argument against a format.

> The source of the message is difficult to read.

Granted, but then, that is the nature of structured documents.

> These are on top of the other obvious 
> allegations that HTML is bloated and insecure.

Easy allegations, not ones that can be founded in reality though. HTML
itself is not bloated, massive text formatting of an email might be.
HTML is not insecure, although a particular implementation might be.

For example, the HTML which comes out of Evolution is little bigger
than the text version (for all the extra formatting), and has never
caused me a serious security problem, nor does it appear it will.

Cheers,

Alex.

1. Ooo, I love an argument
2. Especially about HTML email
3. "Footnotes of Wrath" Copyright Alex Hudson, licenced under the BSD
licence, etc.
___________________________________________________________________

Sheffield Linux User's Group -
http://www.sheflug.co.uk/mailfaq.html

  GNU the choice of a complete generation.