Scribe: Alexey Melnikov
Draft status:
Toplevel (-05) was updated. Has no DISCUSSes left. AD (Murray) is
waiting for the WG.
Haptics: Approved by IESG - AD followup
Suffixes: see later in the session
Standards Tree: WG Last Call ready (not done)
BCP 13 / RFC 6838 rewrite: Started a little bit, Harald found XML source
Harald: who read -07. About 3 people.
MNot: When this started, I was not sure why we need multiple suffixes. I
suppressed my feelings. Most recent discussions made me think that my
feelings were valid. Need some guidelines on what is a useful suffix and
what is not, what is the registration process. What is the exact
semantics?
Martin Duerst: Somewhat agree with MNot, but feeling more positive about
this. Things like +xml are useful, as can be used by a generic XML
processing software.
Martin Thomson (MT): I was less confused about this in the past, but
then I read the conversation on the mailing list. E.g. suffix order
processing confision was concerning. I would rather "the whole media
type is understood before being processed". This is probably not in
agreement with what other people are thinking. Some concerning facts
from https://tess.oconnor.cx/2023/09/polyglots-and-interoperability
presentation, where different processor look/act differently and this
causes security considerations.
Manu: The initial reason for starting this document was the attempt to
register +ld+json. MNot, what concrete things can I do to address your
concerns and get this published?
MNot: I would like to have a section explaining in details use cases for
them. Appropriate and unappropriate use for them.
MT: How comfortably are with suffixes? If you accept that suffixes are
useful, the document is well written. Things like +gzip make me
nervious, does this mean they can be used with any media type?
Manu: MNot/MT - can you help me write this section?
MNot: I didn't think it was a good idea, but I don't have use cases, so
I am not a good person to help with this. But happy to review text.
Happy to put some work in reviewing existing suffixes in the registry
and generally put some work into this.
Harald: Another point is that the current format of the suffixes
registry is a mess. IANA need instructions on how to improve/fix the
registry. This document is a convenient.
Pete Resnick: What started this?
MNot: Maybe just grandfathers existing media types with multiple +,
instead of allowing for them through multiple suffixes?
Manu: Some existing media type existed for 5 years. Some software was
already shipped. Trying to remedy the situation.
Brent Zundel: we (W3C) don't want to break the Internet. We try to do
the right thing. Tell us what is the right thing? What should we do.
MNot: how many media types use multiple +?
Answer: about 5.
Alexey: Why is grandfathering bad? If we add a note saying "don't do
multiple suffixes again".
MNot: Can you change "+" to "-" and ship the code? If not, let's try to
grandfather.
MT: Is suffix a convention? Should we just retire suffixes? Just a
thought.
Martin Duerst: Playing Devil's Advocate for a moment. What exactly are
concerns by MNot/MT?
MNot: See the link above to Tess' "Polyglots don’t lead to
interoperability" article.
Manu: If we can't agree on what is the meaning of suffixes, maybe just
say that "-" and "+" don't mean anything?
Hans-Joerg: Can people a bit of background on this? I though media types
were created for filesystem dispatching to applications (e.g. on
Windows). Other use cases are on the web, where there might be no file.
Does this help?
MNot: I am sorry about possibly wrong advice given by the WG earlier.
Would it be workable if we actually say: "no multiple suffixes are
allowed"? "Would you still love me" :-)?
Manu: We still love you MNot :-). Need to update existing specs that
have already shipped. And existing software can't be updated.
MNot: Can we say 'existing 5 media types are Ok, but use "-" in the
future'. Will this work?
"MNot is trying to negotiate with Manu how much he can get away with".
What is already deployed? What can be changed?
Pete Resnick: Manu, how much screaming and kicking there would be if we
say "don't use + anymore"?
Brent explains how W3C working group work: LD is asleep, but might be
revived in a couple of months. DID is asleep.
MNot: Suggests that we (IETF) work through IETF-to-W3C liason.
Manu: Declare +ld+json the same as -ld+json.
Alexey: Suggest don't create -ld+json as an alias for +ld+json. This
just will cause more interop issues.
Action: ask in W3C whether changing future registrations to -+json
is Ok? MT (as the IETF-to-W3C liaison will introduce Harald to W3C
people)
Magnus Westerlund suggested (not in the room) to merge it with the main
one. Add column with boolean "Is the media type suitable for RTP?".
Close the RTP specific registry and add a link to the global media type
one.
Harald: create a repository, put the XML file he found and ask for
editors.