Re: vger *should* produce quoted-printable

Linus Torvalds (torvalds@cs.helsinki.fi)
Fri, 23 Feb 1996 08:51:27 +0200 (EET)


This is a flame mail, skip this if you don't like those. I'd like to point out
that this flame is not against some person, certainly not any of the persons
quoted here - I'm just reacting to the normal arguments on this matter.

Andreas Kostyrka:
> On Sat, 10 Feb 1996, Alain Knaff wrote:
> > * Many many eons ago, there used to be some machines which used 7 bit
> > characters (in order to pack 5 of these into their 36 bit
> > words). Obviously, those wouldn't support 8 bit characters. Let's call
> > them type A MTA's.
> Interesting theory, but it doesn't matter why, defacto the RFC mandates
> 7-bit messages, or the use of MIME. C'est tout, qu'est important. So
> actually 8-bit MTA's that don't do MIME (8BITMIME extension), but just do
> forward plain 8-bit mails are broken and should be fixed.

[ flame mode on ]

BZZT! Technically correct, but wrong answer.

The fact that the RFC's mandate 7-bit messages is totally irrelevant, because
bad standards _should_ be broken. Adhering to old and outdated standards is a
sure way to keep UNIX unfriendly and backwards: we have to ADAPT to new
circumstances.

I _really_ _really_ hate people who say that everybody must use 7-bit mails
because it's the lowest common denominator. That way only leads to more and
more cludges, and more and more work that is totally unnecessary. Standards are
only good as long as they make sense, and 7-bit data doesn't make sense.

Anybody who has a mail gateway that isn't 8-bit clean these days should be
shot. They _deserve_ to be broken. I do the same thing when I notice that
something is badly broken in the kernel: I try to avoid breaking user programs,
but if something needs to be done it really needs to be done.

Quoted-printable isn't the answer. QP is the question, and the answer is NO.

> Because, before you can urge someone to upgrade to a 8-bit MTA, you have
> to upgrade to our standards (RTFRFC's). So if you want to do 8-bit
> mailing, please follow the standard, and don't solve it the french way.

The problem is that QP is _evil_. And the reason that QP is evil is that it
only helps perpetuate a broken standard. Instead of fixing the standard, QP
cludges around it, and that is why I personally really dislike QP..

Essentially, QP doesn't fix the _reason_ for the problem, it only fixes the
symptoms. It's like giving a cancer patient painkillers and sending him home to
die in peace without even _trying_ to save him first. Sure, painkillers are the
easy answer, but are they the _right_ answer?

> (8-bit plain Emails don't solve all problems, they don't tell you the
> character encoding.)

Now, this is a totally different issue. In this regard I heartily agree with
MIME and all the other mailer extensions: you need to tell the encoding, in
order for people to be able to find out _what_ the data is. It's only partially
a question of character sets: MIME is needed for things like sound, pictures,
video, etc etc etc.

There _should_ be a protocol for telling people that the mail message is in
Arabic or Japanese or ISO-Latin1 or if it's just an audio file. That part is
fine. But mangling 8-bit data down to 7 bits is broken, and has no excuses.
That is why the damn QP stuff is so wrong.

> > In a few years we will see another super-duper "standard" intended to
> > protect messages from QP. When will people understand that a
> Why? Just use RFC-complainant software. Then you have NO problems

Right. You can cludge around everything, and it will work. That isn't the
issue. The issue is that QP is broken as designed, and somebody needs to wake
up and smell the roses.

I really hope that some company like Netscape or Microsoft ("We're so large
that we make our own standards") will make a mailer that sends out 8-bit mail,
and forces people to upgrade their broken mailers. Some standards simply need
to be forgotten about and buried.

Of coure, all the UNIX bigots will point fingers at Microsoft/Netscape and tell
everybody how bad companies they are.. While at the same time wondering why
people think that UNIX is hard to use and why Microsoft is larger than most of
the UNIX vendors combined. Stupid twits.

> [*screammode on*] STANDARDS [*screammode off*] is here the KEYWORD. I
> see, that this may be a bit of a problem for some (I didn't say la
> Francophonie, :) ) people :(

This is an understandable standpoint, but it's not IMHO correct. Standards are
like laws: they are needed and you should follow them.

BUT (and this a HUGE but), there are bad laws (standards) around. And you
should not condone them blindly, you should always question them. And if they
are bad laws (standards) you should fight to make them better. With
non-compliance, if necessary.

That's how the world gets to be a better place. We in western countries tend to
forget that sometimes, because most of our laws _have_ been fought already, and
peple generally think they are good laws. But the internet isn't fully
developed yet, and it has some very bad laws (standards).

[ Flame mode off ]

If you've read this far, you probably can guess that I vote for mail servers
_not_ mangling 8-bit data. I'm probably fighting windmills.. And yes, I have
had to slowly move to the damn QP-aware mailers too, but I can still hope that
there is some "Gandhi of the Internet" out there that is willing to fight for
his rights..

Linus