XSF Discussion - 2017-11-29

Ge0rG: No, no newlines in attributes.

Zash, works for me: >>> etree.fromstring("""<foo a='&#10;'/>""").get("a") '\n'

Didn't we go over whitespace rules recently?

I don’t recall

https://www.w3.org/TR/REC-xml/#AVNormalize

&#10; circumvents that I think

> =xml.parse[[<x y='&#10;'/>]] <x y=' '/>

Indeed

> Note that if the unnormalized attribute value contains a character reference to a white space character other than space (#x20), the normalized value contains the referenced character itself (#xD, #xA or #x9). This contrasts with the case where the unnormalized value contains a white space character (not a reference), which is replaced with a space character (#x20) in the normalized value and also contrasts with the case where the unnormalized value contains an entity reference whose replacement text contains a white space character; being recursively processed, the white space character is replaced with a space character (#x20) in the normalized value.

Zash: but in values you can have them.

replacing the character reference wins over normalizsation to space

> return xml.parse(tostring(xml.parse("<x y='&#10;'/>"))) <x y=' '/>

nice

that should probably not happen

Things broke when we escaped non-space whitespace.

which things?

Things!

Broke interop with M-Link IIRC

even if only applied to attributes?

I don't recall the details

aioxmpp escapes newlines etc. in attributes and haven’t had issues with that yet

jonasw, which xml API does aioxmpp use (if any)?

Flow, for generation, we actually write XML ourselves. for parsing, it’s xml.sax, which is backed by expat in CPython.

(the custom writer implements a SAX interface, but with some goodies which were hard to achieve with xml.sax, see <https://docs.zombofant.net/aioxmpp/devel/api/internal/xml.html#aioxmpp.xml.XMPPXMLGenerator>)

mostly about encoding, namespaces (and prefixes) and strictness.

jonasw, so does aioxmpp do the escaping or does it happen under the hood of the XML API?

let me check, IIRC we invoke some xml.foo method

yeah, xml.sax.saxutils.quoteattr -> libexpat on CPython

wauza, 75% of aioxmpp's SLOCs are tests. repspect jonasw

thank you :)

yeah

edhelas, around?

Guus, is there a way you could give powers to (only) cancel (only) the xeps docker builds?

(you, or any iteam member really)

jonasw: unlikely

pity

too late now anyways

the access control is not very fine-grained

read/write/admin on a repo level, that's all I can find.

okay, nevermind then. I’ll just try to make sure that I only need one push then :)

jonasw: perhaps you can shorten the build duration somehow?

that'd reduce the impact

the PDF build takes ages, I don’t think we can do anything about that

if we had incremental builds, that’d be amazing too, but I think that’s not possible with dockerhub

or maybe even docker in general

isn't the entire concept of docker hub that you base images on eachother?

I don’t think that we can reasonably base on the previous build, can we?

I don't know

Not reasonably, no.

And the idea is that you build up layers, yes, but not generally diffs between multiple versions of the same image.

hm, would write access allow me to break things which I can’t break anyways by pushing to the xeps repository?

as you're an editor, I'm unsure if you shouldn't be allowed to do more damage anyways

there's actually an 'editors' team there

Kev, any good reason why editors shouldn't be part of that editor team on docker hub?

it now has Kev and Sam

Don't think so.

jonasw - can you share your dockerhub accountname?

xsfjonas

sweet

hocus, pokus, pilatus, pas...

thanks :)

no problem

jonasw pong

edhelas, I used "jt" as initials for your change, I hope that’s okay.

ah yeah, that was that :D

is there a place where this jt alias is defined ?

no

unfortunately not

I feel we should maybe move from "initials" to "handles"

initials have low entropy, are prone to collisions, and aren’t used anywhere else in the community, making them hard to resolve

I thought that people that participate will appears in Acknowledgements

not unless you add yourself, I don’t think that’s the editors task. maybe.

(if it is, my apologies)

eheh, we should clarify that :p

I think the Ack seciton is pretty much at the authors discretion

okay

maybe point all those initials to https://xmpp.org/about/xsf/members.html with an anchor ?

ah, this list can be cleaned up as well, which can break links… mhhh

also, you don’t need to be a member to contribute to a XEP

true

edhelas: the members list is much more dynamic than the XEP history

I’m looking forward to the fun of a name change.

eheh

whatever, having pubsub#max_items will now allow us to move forward regarding the PEP support on the servers

:)

sorry for the delay, I forgot this existed and council voted on it :(

I note that the log timestamps on http://logs.xmpp.org/council/2017-11-29#15:55:08 are off by a few minutes

I think five

cc @ iteam

intosi, MattJ maybe?

We'll need to figure out if there is a docker image for ntp, or make one

(only joking)

(I think)

please be.

jonasw: re 0392, "Treat the output as little endian and extract the last-significant 16 bits." - "least-significant" and I'm currently questioning whether it's a signed or an unsigned 16 bit number.

MattJ: right

Or we might've forgotten we actually care about time, and didn't bother installing ntpd ;)

Fixed

intosi, I'm glad you found the time.

dwd: fixed with minutes to spare.

Ge0rG, unsigned

jonasw: or maybe it doesn't matter at all?

unsigned shorts are a real pita in Java :>

Ge0rG, it matters, I think

jonasw: "Divide the value by 65536" - shouldn't that be 65535?

no

we don’t want to reach 1.0

because that’s equal to 0.0

(in the circle)

openntpd seems to be a better choice for just keeping time correct, imho

jonasw: great!

(I got that wrong at first, Marcus pointed it out on standards@)

*Marcel

damn, I got that wrong in the acks

Ge0rG, is yaxim about to get XEP-0392 support? :)

if so, make it so that conversations can steal it :)

Conversations stealing code from Yaxim, that'll be the day

-xep 392

Zash: Consistent Color Generation (Standards Track, Experimental, 2017-11-29) See: https://xmpp.org/extensions/xep-0392.html

jonasw: I'm working on 0392, yes.

wow this XEP

edhelas, you haven’t implemented it yet?? even poezio has it

*cough*

I think it's a well-produced XEP, but really seems like a lot of fuss over something so trivial

mathieui wan't to compare sizes of our implemented XEP list ?

MattJ, I’m sorry :-)

edhelas, no thanks :D

I really don't care that much what colours people show up as

Seems many people do...

mathieui SIMS implemented ? :-°

MattJ: the important thing is not that all clients show the same color but rather that the palette has good readability

the palette stuff blew it up much ✏

(with "palette stuff" I mean mapping to discrete color palettes, like terminal clients do)

Seems weird to me to specify as a protocol tho.

I wonder if it would’ve been more suited for Informational track, but i don’t care :)

(given that we’d want to have this under rather strict change control once it becomes widespread, Standards Track is probably sane)

meanwhile in Movim https://github.com/movim/movim/blob/master/app/helpers/StringHelper.php#L312 :D

the thing is that I'm following the default colors of Material design

so I'm wondering how this will fit with this XEP

edhelas, the XEP allows you to use a palette

there’s a large section on that

the Android avatar palette has eight predefined colors or something

but I would suggest to not do that and instead use the colors generated by the XEP.

jonasw, using a palette won't distribute the colors equaly on that palette though, or will it?

I think this XEP was very well thought out and written, but I can't see it being widely used. It's just too much complexity, and too large to implement easily when instead I could just do what edhelas has

daniel, it will do that only if the palette is itself equally distributed

yes... i'm afraid it won't work for me then. i mean i'll happily give it a try if Ge0rG writes the code

but i really want to keep my color palette

SamWhited, the gist of it is pretty much angle = sha1(text)[:2] / 65536 * 2π; cr = sin(angle); cb = cos(angle); and three formulas to convert cr and cb to rgb. maybe it’d help if there was a single block at the beginning which lists that and save the details for later below?

daniel, why though?

jonasw, why i want to keep my palette?

yes

because i want them to fit into the platform

daniel: +1 :)

daniel maybe we can have the same palette, we are both using Material Design colors :)

daniel, I *assume* that most such palettes are rather equally distributed

if you have the rgb values for them, I’d be happy to check that

jonasw: that might help, but it will still be a very long complicated XEP, and if I can implement it from just that section why is the rest needed?

jonasw, https://github.com/siacs/Conversations/blob/master/src/main/java/eu/siacs/conversations/utils/UIHelper.java#L35

Potential board meeting time. Who is here?

SamWhited, special cases (like if you’re restricted to a palette), precision of the description to avoid ambiguities, and color vision deficiency correction

daniel, thanks

daniel, is that ARGB or ABGR?

argb

I could’ve guessed.

Argh, maybe.

Nyco, Ralphm, MattJ, Martin - does board meet now?

I was guessing not, because you said you couldn't make it

I'm here, but wasn't sure if the mailing list thread flattened it

Wednesdays are very inconvenient. Parking lots, laptops in cars, no heating, and roaming is typically invovled

as we didn't agree on a new time, I thought I'd check in now.

fwiw, 2:30pm-3:00pm GMT every Thursday works for me - as suggested on the mailing list - but there's no point in agreeing between the three of us, without Ralphm and Nyco.

Indeed

I am good with that

nyco, around?

well, we have quorum now - shall we start a meeting now? We can finalize the scheduling once nyco pitches in.

wfm

ralph, do you want to take the lead? You make for a comfortable chair :)

ralphm ^ ?

Well, I thought it would be very weird to have a meeting without an agreed time

First order of business is that we choose a chair and I think we should all be there

I'm perfectly happy to continue with an acting chair for this meeting.

I do agree that choosing a chair should involve all of us

No need for an acting chair, I am still a proper one

So, hoping that nyco will also agree upon the time, let's go for tomorrow 14:30 UTC

Sounds good to me

works for me

I think all of us are on the same DST switchover date, correct? if so, it'd make sense to anchor it to London time - to prevent a sudden shift twice a year?

as UTC is London time now, there's no immediate difference.

Yes, sensible

I'd be very happy to anchor to London

hey, today is my jabiversary: https://mail.jabber.org/pipermail/jdev/1999-November/060711.html :-)

peter: congratulations! :)

heh

peter, nice, in two years we can celebrate your 20th :-)

this is why they call me "Old Man Jabber" ;-)

Oh nice! Congrats indeed!

Wow, suuuper nice, happy jabiversary :D

In case anyone still thought the XSF was the weirdest standards org, the IETF has pressed its claim by publishing 18 versions of the DTLS spec within half an hour.

dwd: how does that compare to BUNDLE with its 42 revisions?

granted, over the last five or six years

peter: congrats, you're now allowed to vote ;)

what, why (@18 DTLS versions)

Who cares what IETF is doing, everyone is doing what Google does, what's good for Google bussiness

</rant>

Google employees participate in IETF standards. :P

google employees seem to ignore MUSTs in RFCs written by other google employees. Seems to be a big company.

dwd: ekr's response is the best...

dwd: wat, link?

zash: https://www.ietf.org/mail-archive/web/ietf/current/msg105575.html

zash: ietf-discuss is like hackernews for... old ppl?

I miss the days when slashdot actually had good news and comments that were insightful

If Brian Carpenter were paying attention to the TLS WG list, he would have known. (Although I still don't understand why ekr felt the need to do that.)