Re: Fwd: New Version Notification for draft-pardue-httpbis-dont-be-clear-00.txt

I also added it to my (growing) list of topics for 7838bis.
There's starting to be enough for 7838bis that we may want
to fold it in there rather than have a stand-alone draft.
I'm interested in helping pull together 7838bis in the next
few months if others are interested, but have a number
of other things I need to work through first.

Best, Erik


On Tue, Mar 16, 2021 at 11:59 AM Ben Schwartz <bemasc@google.com> wrote:

> I think this is probably worthwhile if there's some nontrivially deployed
> parser with this bug.  Otherwise, it doesn't seem necessary.
>
> On Tue, Mar 16, 2021 at 3:42 AM Julian Reschke <julian.reschke@gmx.de>
> wrote:
>
>> Am 16.03.2021 um 01:47 schrieb Lucas Pardue:
>> > Hello HTTP WG,
>> >
>> > Anthony and I found a little Alt-Svc "what if" case where some naive
>> > parsing code might stumble when trying to process keyword "clear" or an
>> > ALPN protocol ID of "clear". The ALPN space is large enough that the
>>
>> Not "naive", but really broken, right?
>>
>> > chances are small, but we wrote a little I-D (forwarded) that would
>> > reserve the sequence so that nobody might do this unwittingly.
>> >
>> > Thoughts?
>> >
>> > Maybe the odds of actual problems are so small that doing nothing is ok.
>> > Or maybe this is a potential improvement to bank for some type of RFC
>> > 7838bis. Either way, the problem seems well-formed enough to document in
>> > order to gather feedback or criticism :)
>> > ...
>>
>> If we did a 7838bis, we *could* add a warning that using a broken parser
>> is bad. The problem is that people who actually read the spec whould
>> probably know this already.
>>
>> Best regards, Julian
>>
>>