Mailing List Archive


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

Re: [tlug] Changing subject in thread.



Benjamin Tayehanpour writes:

Re: rich-text standards vs. MTA implementation

 > Ah, so there is no actual formal standard for enriched text in
 > e-mail?

That's an inaccurate characterization.  There are (at least[1]) two of
them: text/enriched and text/html.  It's a SMOP to implement both, and
the mechanisms that enable MUAs to determine which is which are also
well-defined (MIME).  The problem is quality of implementations.

 > Spam messages, however, often contains *extreme* formatting, with
 > embedded images and whatnot.

Messages formatted by Outhouse also often contain extreme formatting.
Some correspondents even configure nicely patterned letterheads and
backgrounds (== embedded images).  But this can't be configured by
recipient.

 > In many webmail clients, sending HTML e-mails is the standard
 > behaviour, so one could also consider *that* an implementation or
 > configuration failure, also frequently occurring.

I do consider that a configuration failure, and the fact that none of
the popular MUAs allow you per-addressee configuration is a design
bug.  But I can't do anything about users' MUAs (well, I can write
Internet-drafts, but that takes about 10 years to have an effect, let
alone promotion to RFC).

 > [In Uganda], bandwidth is *really* at a premium, with some
 > operators bleeding you dry if you step outside the established
 > boundaries for your purchased bundle. I would expect a plain-text
 > policy in their mailing list, but, interestingly, almost everyone
 > seem to send HTML e-mails regardless of any actual formatting
 > taking place in them. I don't know what to make of that.

grep the spool for '^(User-Agent|X-Mailer)' and ask again. :-)
That is, you don't expect many of them to be using Mutt, do you?

Re: HTML bloat.

 > > Of course it does.  The HTML part (and GIF background, but I
 > > digress) bloat my spools and archives dramatically.
 > 
 > I don't know your definition of "dramatically", but I just made
 > some size comparisons on gzipped plain-text vs. gzipped HTML, and
 > there was a 2.5% difference.

Your experimental result is of course correct, but irrelevant.  By
your own suggestion, those who want HTML should get it and those who
want text/plain should get it.  In the current implementation of
Internet mail, that means that both are sent to each subscriber as a
MIME multipart/alternative message.  So that's more than 102.5% bloat
(due to the additional overhead of MIME), because I don't need or want
the HTML at all.

Also, note the reference to mail spools.  Because of upstream policy,
the spam (overwhelmingly HTML) gets marked, not rejected, and so ends
up on disk.  Spools here are not stored in gzipped form, and I don't
know anyone who does.

 > Assuming (generously) that 10% of all messages are enriched,

You may wish to make a note to yourself to avoid such parenthetical
glosses in the future.  Of the 376 messages in my tlug/CURRENT folder,
280 have Content-Type text/plain, and 96 have multipart/mixed.[2]  Of
the latter, 93 contain a Content-Type multipart/alternative with two
parts, of Content-Types text/plain and text/html.  So we can conclude
that even in the "hostile" environment of TLUG, a quarter of messages
contain HTML, and around 20% of that folder is unwanted bloat.

Presumably this would rise dramatically if HTML were encouraged.

 > As you say, it's not rational. Unlike you I haven't investigated the
 > matter thoroughly enough to give you an accurate figure expressed in
 > parts-per-hundred, but I'm fairly convinced that a notable share of
 > the *genuine* messages in my inbox is HTML.

Agreed.  Any of my university folders would be way above 25%, and many
of the messages contain *only* HTML.

That's not the point of "smells like spam," though.  If I were handed
an unopened HTML message from my mail spool (*not* my mail folders
which are post-filter), and forced to bet 10000 yen on spam or ham at
even odds, the bet would be "spam".  How about you?  (Denominate the
bet in euros if you prefer.)

 > With a decent formalised standard for enrichment, regexp-based
 > filters would work.

[...]

 > > HTML makes it far easier to insert viruses and malware.
 > 
 > This would be a good point if I meant HTML in the first place. As it
 > was, I'm enquiring about the *idea* of rich-text in e-mails. Suppose
 > an RFC would be drafted which outlines an economic, easily-parsed,
 > easily-filtered method to insert orthographic emphasis in e-mail.

It might be drafted, but even if it made it to RFC status, it would
probably have little effect.  Emacs/Gnus would probably implement it,
and Mutt might.  Everybody else would ignore it in favor of HTML
*because* HTML allows embedded images, audio, and video (and
executables :-P ), not to mention CSS and external bodies.  These are
commonly used features in private mail and many channels.

I will note that if you pass many, perhaps most, of the messages on
the Python-Dev list through a ReStructuredText formatter you'll get
very pretty rendering via HTML or LaTeX.  I don't know anybody who
bothers, though. :-)  Nevertheless, there's a sliver of hope....

Re: random misreadings.

 > That was meant as a counter to the argument that orthography is the
 > only important thing to consider when conveying thoughts by script.

If Daniel decided to take *strong* exception to that interpretation of
his post, I'd be with him 100%.  AIUI, he was arguing that accurate
expression of semantics is a necessary condition for accurate
communication, using precisely correct syntax in the plain text
version is the *first* thing to consider, and *usually* sufficient
(especially on a relatively technical list like TLUG).  That's quite
different from your straw man.

Re: conforming to custom.

 > Customs are a fine thing to have, as long as you remember the
 > reasoning behind them. One way of doing so is to have them challenged
 > once in a while :)

I have to disagree.  One of my favorite quotes is from one of Alexei
Panshin's juvenile novels, where he wrote "Civilization consists in
increasing the number of things one can do without thinking, allowing
the mind to focus on more important matters."[3]  Customs are an
important construct for civilization, and one should be abandoned only
when a clearly superior alternative is offered.  Superiority is decided
by the group, not any individual, of course.

 > If a rule (and I'm not necessarily referring to this one
 > specifically) is made superfluous due to technology marching on,
 > keeping the rule would be pointless. Wouldn't you agree?

No.  Customs are good for us.  I *would* agree to a modified version,
that it's pointless to *impose* the rule on *new* communities.  I
would expect that many would choose rich text even if their MUAs
offered a fair choice (which they don't, cf. Uganda LUG).  But some
might prefer plain text even so.

 > Your analogy is flawed. In your convenience store scenario, I am
 > decidedly at a disadvantage, since I am a lost gaijin and thus "in the
 > wrong" from the get-go. The interaction is a simple duologue with the
 > sole purpose of acquiring a piece of information from the store clerk.
 > In a mailing list, however, I'm in an eternal *dialogue* with my
 > peers.

The analogy is somewhat flawed, but that's a property of analogies.
Your rebuttal is inaccurate on two points.  It's not eternal, and it's
not a dialogue.  (Unless you mean in the technical sense of a
particular kind of theatrical performance.  Plato's Dialogues were for
the benefit of an external audience, of course, but not even Socrates
was aware of us.)  Both points are important; it's not "eternal"
because you can go elsewhere, and the preferences of the lurkers
matters.

 > If everything works as it should, the exchanges (and benefits)
 > should be mutual.

Be careful with your modalities.  The exchanges *will* be mutually
beneficial, but that's because of a generalized Invisible Hand guiding
voluntary exchanges.  But there's no reason we *should* adapt the list
to your preferences, even if "you" outnumber "us".  If you do, there's
nothing stopping you from establishing your own list and letting
Darwin do his thing.

 > While I obviously have no right to *defy* currently held standards,
 > I do have the same right as anyone else to ask about and discuss
 > them. No more than anyone else, naturally, but also no less.

Sure, but you also have "obviously" over-exercised your right here (it
doesn't help the case the Bruno has done so as well).  That's why we
(me too) are about to get ourselves moderated.  Better send this now! :-)

So, cya.  I'm done.

Footnotes: 
[1]  I don't know if RTF is registered as a text subtype.  OTOH the
Emacs/Gnus MUA can fontify such arcane types as application/emacs-lisp
(but that's not IANA-registered).

[2]  This is due to Mailman's algorithm for adding footers to
multipart messages.

[3]  The "thing done without thought" in the novel was to survive a
rigged duel with impeccable style.

[4]  For a really horrifying example, consider the "rich kid's
disease", polio.  It turns out that polio likes to attack 7 and 8 year
olds because antibodies received from their mothers take months to
become ineffective so they're protected in the hospital, and the next
time they are regularly exposed to virus-rich environments is in
elementary school.  Rich kids who live in hygeinic environments never
build up antibodies, but poor kids do.  So ....


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links