XSF Discussion - 2017-03-23

Github doesn't have a way to diff between the old version of a PR and the new version, does it?

You can diff between arbitrary revisions if you figure out the URL format

Not sure exactly what you mean

Tobias: it does but is very well hidden

Guus, a student did a PR, i provided feedback, he updated the PR based on the feedback. now i want a diff between the changes he made

i remember some people working around by only adding commits and squashing them in the end after the review

Tobias: that is my prefered workflow

ah, he already squashed? Not sure if you can look at the previous commit - didn't that disappear after the history got rewritten by the squash?

i'm just more used to the gerrit work flow

note that i'm somehwat of a git novice. Listen to Flow.

Guus, not really...the feedback changes basically were amended to the original commit

MattJ don't make eye-contact, and remain calm. Do not run.

MattJ: it's safe here. We have reflog.

Guus: you could try https://github.com/$user/$repository/compare/$commitA..$commitB

if you still have the commit IDs somewhere

jonasw: i will not, but perhaps Tobias would like to.

uh, I accidentally killed my muc history, sorry :)

jonasw, ta...will give that a shot :)

jonasw, sadly..the old commit is gone

Tobias: :(

then I’m afraid there’s no way

luckily the PR isn't that large

Orphaned commits aren't pushed, but if you pulled it before you should still have it.

hmm i just noticed that https://xml2rfc.tools.ietf.org/ link to a XSF feed at http://www.xmpp.org/extensions/rss.xml which doesn't appear to get updated any more

same for http://www.xmpp.org/extensions/refs/, although I'm not sure what is supposed to be at this location

can we create a github issue, trello card, $note-somewhere, reminding us to fix this so that we can forget about it? :)

Hi, small pubsub question: can an entity with "publish-only" affiliation retrieve its own items? It can for other ones, but it's not clear for its own. It seems this affiliation can fulfil a feature I need.

it *can't* for other ones

Flow: please create a new issue here: https://github.com/xsf/xmpp.org/issues

I eventually start annoying people to fix those

ralphm: hi, you probably can answer my question ^ :)

goffi: I would say it would not be allowed to retrieve any items.

That would include its own.

intosi: It has been a couple of weeks - https://github.com/xsf/xmpp.org/issues/243

Indeed.

:)

Tobias, can https://github.com/xsf/xmpp.org/issues/204 be closed?

yes

tx

https://dev.ssllabs.com/ssltest/analyze.html?d=xmpp.org&s=2001%3a4800%3a7810%3a512%3a14e1%3ab81%3aff05%3a8bb&hideResults=on&latest

Thanks

So, who has anything nice to say about Ryan Eatmon, Jeremie Miller, Julian Missig, Thomas Muldowney and/or Dave Smith? In particular, why they're emeritus members of the XMPP Standards Foundation?

I don't know any of them, but feel that we should give them a bit more credit

well, Jeremie Miller started the whole mess IIRC

I recognize one of those names.

Are the others from before my time?

I'd love that for Jeremie. "Jeremie Miller - he started this whole mess."

Zash, who of them isn't before from your time?

Guus, yeha..don't quote me on that ;)

but, if anyone can write a paragraph on any one of them, please add it here: https://github.com/xsf/xmpp.org/issues/275

Tobias: right

Kev: with your iteam hat on, are you okay with the maintenance burden imposed by https://op-co.de/tmp/deprecation-mail.txt ?

Or is the handling of such PRs something that Guus can take care of until the website situation has been clarified?

Ge0rG, i don't see any maintainance burden for the iteam in there

(why me?)

because you've been most active on github from all...that's the reward you receive for that ;)

Guus: you are the only one I know who's actively maintaining the website. But this is probably due to my ignorance of all the others doing hard work, sorry.

we need more naming and shaming!

https://github.com/orgs/xsf/people

there

those all have access, i think.

Basically I need a volunteer to merge the PRs, I can take care of preparing / reviewing them

I shall be happy to push the bright green button

Guus: thanks very much!

also: where's the blog post(s)?!

Guus: I'll make one with the contents of

https://op-co.de/tmp/deprecation-mail.txt

works for me :)

to make it even more "official": https://linuxfr.org/news/rencontre-xmpp-jabber-par-jabberfr-mardi-28-mars-2017-a-19-h-a-paris

sorry, not English

ah, cool

nyco, can I tempt you to maintain the list that is now on https://wiki.xmpp.org/web/Meetups on our website instead?

The alternative is that we delete the existing but outdated event page: https://github.com/xsf/xmpp.org/issues/276

(which I already did, but the PR has not been merged yet)

Guus, what's wrong with the maintenance? I did that page, and it looks good to me, so please guide me ;-)

nyco, You are referring to the wiki page, I assume?

this page lists old information: https://xmpp.org/community/events.html

"Summit 19 will be held once again in Brussels in February 2016"

We should either update it, or remove it.

the wiki page is pretty similar - so, I was thinking that instead of having both pages, we'd combine everything on https://xmpp.org/community/events.html

that, or delete that page.

(the Lanyrd link shows feb 1st, 2013, as the most uptodate entry)

whatever is best...

I guess the website has more visibility?

hey, can I blog on xmpp.org about the French/Paris meetup?

yes to all

generally, you may want to go to Reddit website and find the XMPP subreddit, and there you might wanna upvote some articles and comments, and add your positive/insightful comments (not linking directly, because don't wanna generate fake traffic that will be penalised)

Guus: https://github.com/xsf/xmpp.org/pull/282

Ge0rg: did you testrun this?

I think the title might be to long for comfort

Guus: negative.

Guus: haven't looked into rendering xmpp.org locally yet, merely pandocced it once

Guus: do you have ideas for a catchier title?

"XMPP Software Developers: Action Required" is the mail subject

maybe "New XMPP Software Listing Rules"? ✏

Never going to give you up .. oh wait yes we are

fyi: vagrant up && vagrant ssh then: cd /vagrant && make devserver website on http://localhost:8000

But ... a static site...

Zash: try and see.

bash: vagrant: command not found

it looks like `./develop_server.sh start` launches that server

apt-get install vagrant ?

Guus: Permission denied

I'm not someone who's going to install things like that

works fine without vagrant :)

(assuming you install pelican and its dependencies…)

Guus: changed title

Ge0rG: I added screenshots pre title-change

see comment

jonasw: obviously, but I'm not someone that's going to install all tools for each project on my laptop without some virtualization. Keeps me sane. :)

it’s not a trivial trade-off

Zash: the entire point of installing that is that you can then install other stuff in virtual machines, instead of directly on your laptop :)

jonasw: true, but it works for me.

I'm a freelancer, dancing around several customers, each with their own projects and project dependencies.

at some point, 3 different versions of the same database software gets annoying :)

I’m frequently running out of disk space. The only thing I find which is reclaimable are the dozens of vagrant-based VMs.

hah :)

yeah, but I gladly pay for some extra disk space. That's the cheapest of resources

(which is the trade-off: isolation vs. disk space)

(docker might actually be a better solution for many of my problems, but I've not looked into that good enough)

it may be, but it’s still expensive…

docker doesn’t help much with that

Guus: commented on your comments :P

"I'm calling you to tell you that I sent you an email"

the new title fits nicely, btw

Ge0rg: you removed a blank line in your second commit, was that intentional?

Guus: > and merged the first two paragraphs to improve the preview.

probably yes

ah, sorry

kk

Guus: yes, it moves "the XSF Board has decided that all implementations have to reapply once per year" into the preview

new screenshots

do we have a more fitting category than "misc"?

jonasw: pondered that as well

you can simply make one up

but make one that's re-usable

XSF Organisational exists, but I'm not sure if it's fitting

Guus: nope

I pondered about that as well. Maybe "Software"?

But it would be the first one.

I'd also like to have something to file "Easy XMPP" under

I don't have a preference

"State of the Union"

"Federation" is a nice word

We need to steal it back from the OStatus people

"State of the Federation"?

Federation of the State?

Sounds like Star Trek.

State of the State?

Federated State of the Union?

https://github.com/xsf/xmpp.org/pull/283

please someone to review?

Guus: I'd like to get the mail out, and I proclaim that 282 is now good enough[tm]

+1

Ge0rG: merged. It should pop up on the website anywhere in the next few weeks.

(hours, likely)

nyco, reviewing

Guus: thanks very much! :)

Guus, thx

nyco: title is just a tad to long

Mail sent. Another day well spent.

Thanks Ge0rg

I can't even imagine how some entries landed on that list... like Apple Messages.

Flow: rush to the githubs! Your library is in danger!

Ge0rG, shall I do the trial run?

Ge0rG: you might not like it, but it supports XMPP, and I know folks who actually prefer it.

jonasw: your README is b0rked: $ ./update-entry clients.json yaxim zsh: no such file or directory: ./update-entry

hmm json

Ge0rG: meh.

does the timestamp require a hour and minute, and if so, why?

Flow: now is too late to complain

not complaining, just asking

are there plans to add a "last release" column?

intosi: is it the default iOS app? How do you add XMPP to it?

Flow: that was discussed, but it was noted that that might be a false classifier: some software that has not been released in ages is pretty stable.

or how about adding an mail address field, which will receive mails a month before expirey

Flow: I like the email idea.

Flow: naaah

Flow: we want active entries, not people who barely care enough.

Ge0rG: I shall forget to update my entries, even if I'm one of the people managing the site where they're listed.

Ge0rG: it's default on OS X. And you add an account by going to the Messages menu, then Add Account, then select Jabber.

* select Other, then select Jabber.

Haven't checked on iOS.

I'd like to see more info on that page, but "last release date" might not be the most important one.

https://github.com/xsf/xmpp.org/pull/284 is correct, jonasw?

Flow, you might as well drop the 'SE' ?

Guus: why?

it's hardly relevant? Hey, keep it if you want :)

it don't think Smack would work on Java ME (if that's still a thing)

What about EE, does that still exist?

jigsaw ftw!

exists, sure, is it a thing, don't care

Guus: what would be better? collect all the software renewals in one PR or make individual PRs per app?

don't really care.

Flow: not sure if your timestamp needs a zone?

jonasw?

example didn't had one IIRC

kk

I'll squash and merge

sorry, was AFK

Can we make Travis check the format of those JSON files?

Guus: if it builds the website, it checks the JSON files

which would safeguard against typo's?

hmm, entries with malformatted timestamps are simply omitted and do not raise a hard error currently

Flow: didn’t give it much thought which information to include in the timestamp. it is in UTC though (without the explicit Z) because I didn’t want to open the timezone can of worms

jonasw: "Use the tool as described in the previous section to perform a renewal (this will sort the list correctly to minimize future diffs)" is a lie, it doesn't sort :(

it does sort the keys

it doesn’t sort the applications, that’s true

jonasw: that's the opposite of what I'd like to have actually ;)

I'd rather have "name" first

and the applications sorted

Ge0rG: I can’t force the json module to do a specific key sorting

which means that without sorted keys, we would end up with random sorts on each save

which is very bad

Yay for python json

it’s based on a security feature

jonasw: but you could sort the applications in the array by name, couldn't you? :D

yes, that’s possible

jonasw, can’t you use PYTHONHASHSEED?

to fix the dict key order

mathieui: and find a hash seed which happens to do what we want?

interesting hack.

:D

requries everyone to set PYTHONHASHSEED though ;-)

(you can’t set it from within python)

Ge0rG: I’ll make a PR which will make a nasty diff and sorts the entries, and then we can discuss there if we want to do that

jonasw: +1

damn, I was too slow in adding Bruno, now it's a new PR

ah, nice, it doesn’t even produce a diff

You snooze, you loose, Ge0rG

Guus: jonasw is at fault, I was following the README and got confused about the non-sorting

blame shifting!

jonasw: incorrect documentation :P

I have a hard time having a serious discussion with someone that just added a software entry named "bruno the jabber bear"

Ge0rG, Guus: https://github.com/xsf/xmpp.org/pull/288

jonasw: Ge0rg appears to be making up info categories, is that ocrrect?

Guus: what about the person that is maintaining Bruno the Jabber Bear for over four years now?

jonasw: you need to change that newline in data/libraries.json?

Guus: it is what happens when you run the tool on the file

k

Guus: you mean the ones in clients.json? I think we need to rework them all, categorically

the tool doesn’t put a newline on the end, I ran it to canonicalise

ack, Ge0rG

ideally, we would have them worked out in such a way that we can automatically split the tables on them as has been proposed in https://github.com/xsf/xmpp.org/pull/193

jonasw: yes please. We can work together on a canonic list of platforms

Android, iOS, Linux, Windows, MacOS

anything else?

heh. No love for windows phones :P

"Browser" "Console" "IBM i" "Linux" "Mobile" "Mobile (Android)" "Mobile (Android, Blackberry (BBOS), Nokia Symbian S40" "Mobile (Android, iOS)" "Mobile (BlackBerry)" "Mobile (iOS)" "OSX" "S60 and Asha)" "Text-Mode" "Web" "Windows"

wow.

right, browsers probably make sense

do we have to have mobile permutations?

splitting for common mobile OSes probably makes sense

just allow multiple OS categories

we do

My special favorite is the unterminated "("

Ge0rG, That had my OCD kick in.

sorry, my fault

")"

that was my code foolishly splitting on / when creating the entries. the "Mobile (Android, Blackberry (BBOS), Nokia Symbian S40" belongs to the "S60 and Asha)" one later

I’m preparing a cleanup now

jonasw: good dea to properly refactor it now.

and then have a keen eye on the PRs

dwd, just open your scheme handbook for medication

jonasw: we ideally have some sort of script that checks for this, which we can hook up on travis or another github check.

can do

what's the current name of OS X? "macOS"?

macOS

can we agree on the list of platforms first though?

jonasw: yes please

suggestion: [ "Android", "iOS", "MacOS", "Linux", "Windows", "Web", ]

jonasw, happy to...if new platform comes up we can always issue another PR

that's macOS with small m..otherwise looks fine

although, there are those blackberries

I'd go with "Mobile", and "Mobile (iOS only)" "Mobile (Android only)"

Mobile is implicit with Android and iOS

Guus: I’m doing this with prospect that we might at some point automatically split the table by platform. Having Mobile (iOS) and Mobile (Android) is annoying there.

(and another Mobile which implies both)

jonasw: "MacOS" -> "macOS"

Guus: -1

jonasw: maybe "Browser" instead of "Web"?

discuss!

and what to do with "Console" and "Text-Mode"?

let's keep it simple?

console, web, ios, android

console isn’t a platform

it’s a frontend

"misc"

"other"

no

nyco: is your console stuff on MS-DOS, Windows, Linux or whatever runs on my router?

jonasw: all UI is a front-end.

"unrelevent"

jonasw: "Windows" is a frontend to NTOSKRNL

KISS

just allow free tagging for OS values and we try to normalize it on demand as a community?

another question, do we want this also for servers? if so, we need to add "*BSD" and "Solaris" to the list :)

jonasw, "POSIX", and a massive argument as to whether that strictly covers Linux or not.

Linux is a Unix, so maybe we need a generic Unix?

fun!

Ge0rG, not so fast there

:P

jonasw: also rename 'info' into 'platform'

it will conviniently move it below 'name'

Ge0rG: in the tooling? that’s complicated, it doesn’t apply to libraries

Is Linux a Unix? I don't think that it is; maybe some individual distros pass the Unix specification?

You guys made Kev run.

macOS is a certified UNIX.

seriously though, what to do about Console / Text-Mode?

jonasw, Nothing?

dwd: Nothing as in "keep it" or Nothing as in "drop it"?

what about non-IM clients?

I'm slightly for "drop"

jonasw, It's not a platform, certainly.

jonasw, So drop it.

then we end up with some clients without platformbs

because they only specified "Console / Text-Mode" (e.g. Finch and GNU Freetalk"

like bots and such

s/"/)/

ugh

jonasw: remove non-matching items from non-renewed software, let authors sort out

clever

Ge0rG: I’d rather leave it as it is and we sort them out when the software is getting renewed

but I’m going to canonicalise those for which it makes sense now

jonasw: I'm still for renaming "info" into "platform", because programming languages are platforms as well

Ge0rG: you could’ve brought that up before it was merged

jonasw: Guus is too fast.

the PR was hanging there for weeks

well, at least one

(I think)

I WANDERED OFF!

hm, maybe only two days

jonasw: I'm not in the habit of opening and reading random PRs

"the plans were there in the local planning department on Alpha Centauri"

https://twitter.com/Xabber_XMPP/status/844865634672435200 - awesome!

I wonder whether daniel knows about that

Holger, daniel: engage your lawyers! https://twitter.com/Xabber_XMPP/status/844865634672435200

Guus: https://github.com/xsf/xmpp.org/pull/289

jonasw: neither "Blackberry" nor "Nokia Symbian" make sense.

i have no idea

hmm I wonder if kontalk and conversations.im got a similar letter

it was written there in Mobile (…)

jonasw: I will definately forget to do the manual checks at some point. Please make travis fail when desired.

jonasw: it's slightly similar to writing "Unix"

Guus: the issue is that we cannot agree on what is desired

jonasw: I propose "other mobile" for platforms that lost the mobile platform war

jonasw: that does not take away the fact that travis should fail on a to-be-determined list

okay

can you do linting of the json files too?

will do

💕

intosi: it seems you're right, thanks for the reply. That's unfortunate, I would like that an entity can write/retrieve or or more item(s) without being able to get any other one

jonasw: "info" -> "platform"? :D

Ge0rG: it’s on the way… I only have two hands

one step after the other

jonasw: yay! :)

wait Ge0rG you have to let the german govt know if you start a server on your machine there? or I read it wrong?

moparisthebest: if you are running a commercial public service

Don't forget to register your blog with the ministry of truth.

so, I ran forums (with private messaging?) and still an IRC server on a hetzner server in germany since 2006, I'm breaking german law since I never contacted anyone?

moparisthebest: only if you have ads on it, or advertise yourself as a professional somethingsomething

the forum had ads, the irc server does not

moparisthebest: the forum qualifies then

what's the purpose for this anyway? requiring backdoors or?

moparisthebest: I'll let the authorities know, expect an extradition request soon.

ha I guess so :)

Just put up a sign that says it's verboten for Germans to use your service

maybe huge german hosting companies like hetzner might let foreigners know this...

moparisthebest: the reason is to have a list of companies providing telecommunication services to the public, to better protect people's privacy and to allow for lawful interception

those last 2 items are completely contradictory, but whatever :)

add some doublethink in there and it's fine

Ge0rG: Is this new?

Or is this the data retention thing resurfacing?

Not new. We have always bin at war with Eurasia

seriously with this definition what *doesn't* constitute as a 'telecommunication service' ?

Zash: not sure, maybe ten years or so. Not much to do with data retention.

I think a static html page without javascript might, but that's communication in one direction to multiple people even

there is also a public directory of such telco providers, with proper postal addresses given (probably so you can sue them)

daniel: did you already saw the Android O preview?

creepy, so where could you properly host an xmpp server? sealand?

define properly

define your threat model

Flow, i saw the announcment. https://twitter.com/iNPUTmice/status/844257001668513793

yep exaclty, note that they also did something similar in the Android M preview

Zash, properly as being able to just start a server and not register with the ministry of truth first

and later added the whitelist for battery optimizations

in the final release

i'm still a bit unsure if the permission to ignore battery optimizations might not include that as well

otherwise how are the playservices supposed to run?

moparisthebest: I'm not aware of any such requirement in .se. Also pretty sure that anything self-hosted is fine.

daniel: I hope so, but I also wouldn't be suprised if google made it an system-app exclusive feature

Android O looks like Android 0.

Android Zero

abwarten und tee trinken

daniel: do you know from your head what is needed to make a foreground notification that's slightly greyed out and not displayed together with actual message notifications? (I don't even know how to properly word what I mean, sigh)

Ge0rG, low priority

daniel: thanks!

daniel: you could create a Conversations wallpaper service!

Guus, Ge0rG: https://github.com/xsf/xmpp.org/pull/290

jonasw: platforms += ["Other"]?

Ge0rG: really? let’s see how it plays out without that.

would be okay with replacing "Other Mobile" with "Other" though

jonasw: might work well enough as well

jonasw: are those platforms.json for clients only or also for servers and libs?

not intended to be used for libs, may be used for servers in the future

but currently that’s not enforced

jonasw: I imagine a common platforms json for server and client is okay, and a separate one for libs

OTOH, you don't really often need to filter libs by platform.

I don’t think we need one for libs, right

for clients, there is immediate benefit.

Is the clients.json also hosted somewhere on xmpp.org? I could use it to automatically generate the list in easy-xmpp-invitation

it is not, afaik

but should be possible

BTW, it would be nice to also have a description field in the json, though it should be named differently, to be after "name""

Ge0rG: it obviously is hosted here: https://raw.githubusercontent.com/xsf/xmpp.org/master/data/clients.json

Guus: thanks

Ge0rG: updated

Ge0rG: it wouldn’t be shown on the website then though, right?

Guus: I’m not that good with travis, also I don’t know the environment. However, running ./data/lint-list.py clients.json ./data/lint-list.py servers.json ./data/lint-list.py libraries.json is what you wanted me to make possible. Those will fail with non-zero exit codes when there are errors.

jonasw: travis executes: make html

does it make sense to hook it into that?

Guus: I’d rather add additional commands

jonasw: "it"?

Ge0rG: the description

jonasw: would it be hard to add?

Ge0rG: no, not particularly, but I wonder if that’s the right place.

and if it will be well maintained if it isn’t shown on the website, and what’s next

also, wasn’t someone else working on a registry of software?

or rather, a way for software to publish their manifests

then we could simply add a url_manifest or something like that

jonasw: if I read https://docs.travis-ci.com/user/customizing-the-build/ correctly, you can simply add items to the script directive in .travis

Guus: do you know if python3 is available in the travis build?

https://docs.travis-ci.com/user/languages/python/

Guus: this will make different builds with different versions of python.

the default appears to be 2.7

I have to adapt the linter to work with that

which is not trivial because the .casefold() method used for sorting requires python 3.3+

it is all chinese to me. What happens if we make the entire thing run in python 3.3?

Guus: not sure if pelican survives that. at least it probably makes the environment on the server more different from the environment in travis

I’ll try to figure out a way to make travis run the linting with python 3.3+ and the website build with 2.7

I'm not sure what version of python is running on the server in the first place?

I don’t know either, but pelican still defaults to 2.7 afaik

It's a Debian system, so 2.7.

By default, that is.

I didn't dare run Pelican with anything else.

does anyone know if dino has a muc? or like, any way to find out what xeps it supports other than source code spelunking? https://github.com/dino/dino ✏

A muc, or muc support?

moparistthebest: we don't have a muc yet, but setting one up is on the todo. And for the supported xeps, they are roughly "listed" here https://github.com/dino/dino/tree/master/xmpp-vala/src/module/xep , although some are only partially implemented

no stream management?

hi lskdjf, you are one of the devs?

yea I actually meant 'do they have a muc' but I would have also been interested in if they had xep support

jonasw: no, not yet

thanks much lskdjf

moparisthebest: star https://github.com/dino/dino/issues/13 :P

moparisthebest: muc xep support exists, yes

Ge0rG, yes I am

star or thumbs up? I only see thumbs up

moparisthebest: star up and thumb it!

I thumbed it

lskdjf: I wanted to pitch https://wiki.xmpp.org/web/Easy_XMPP to you :D

never heard of vala before, looks kind of java-y which I like, will watch closely :)

moparisthebest: The point of Vala is avoiding C in the GObject/GTK+/GNOME world. The syntax is built around the GObject model and it compiles to C code.

moparisthebest: As far as I read, vala is mainly ment to be c#-like (which again is similar to java...). But it's really nice. I always thought of UI-development being a horror, but I started liking it, vala integrates GTK really nicely.

moparisthebest: Other than that it's really just yet another totally boring Java/C#-like language yes :-)

I absolutely *love* the idea of languages compiling to C, I've seen a couple like that

a totally boring C#-like GTK-only language.

it's pretty ingenious, then you get compiler and great optimization for free

moparisthebest, No, you don't.

moparisthebest, You get a compiler, sure, but you could have got that by writing an LLVM frontend.

Doesn't GCC still have better optimizations than LLVM & co?

moparisthebest, You don't get as good optimization, because you have to have machine generated C in the middle, and that loses both semantics and probably injects a lot of boilerplate.

it seems like when a new archictecture comes out a C compiler comes with it at minimum, not necessarily llvm output

But who cares, the CPU will just re-optimize the machine code anyways.

Zash, Different level. It's not like you can run badly written C and it'll miraculously recompile it.

Zash, It's more that some of the ASM hand-optimization is done by the CPU these days.

dwd, I'm not convinced a compiler author could output good llvm but not good C

moparisthebest, More scope, I think, for optimization at the transcoding stage.

I'm not sure how llvm handles that, if you can have 'optimal' llvm code, how is it optimal across platforms where there are different optimal ways to do things?

but also it locks you to llvm which doesn't support everything, idk, I still think in theory I like the C idea better, practice may be different

hello everyone, I'm planning to work on some XEPs in the upcoming days. I've already done some work on the OMEMO XEP based on the discussions that we had during the FOSDEM Summit of this year.

what is the best way to submit it, should I do a PR ?

edhelas: Yes please!

perfect :) I'll do that asap

also I'm (finally) planning to work on the Bookmark XEP

SamWhited: are you watching the xeps repo or shall one notify the editors team directly when one issues a PR?

jonasw: It will email me when someone opens a PR

okay thanks

(yes, watching, what you said)

lskdjf: small patch incoming :)

SamWhited, because I'm changing how OMEMO bundles are published and retrieved, should I bump the urn:xmpp:omemo to :1 ?

it seems that no clients is currently using urn:xmpp:omemo:0 for now

moparisthebest: source-to-source is often worse than source-to-IR

edhelas: Sounds like that's a breaking change, so I'd think so, but I'm not sure. You could chat with daniel about the change maybe

afaik andy is going to bump to omemo:1 pretty soon anyway

edhelas: I think you should have coordinated your changes with the authors

it's not done yet :)

just wondering so I can actually coordinate

but if they are small enough, just issuing an PR may be also without that mutch overhead

i've heard that there is a secret MUC where all the OMEMO devs idle around

but maybe it's just a rumor

they are all encrypted :(

so how far is the movim implementation of omemo?

for now on pause, that's why I'd like to work on the XEP first

I think that I'll work on it in a month of two (got some other things to implement first)

ok but if i remember the thing right about bundles, this was only a cosmetic change, nothing which would you hold back in an implementation

or is there more you want to change?

mostly how the bundles are published for no

https://edhelas.movim.eu/0384.xml

SamWhited, I propose this as a starting point: https://github.com/xsf/xmpp.org/pull/294

arc, please add PyCon stuff, sorry...

nyco: did you mean to ping Guus?

oh, as well, you're right... and you jonasw as well ;-)

ping all, in fact ;-)

how do i find out if im still in a room?

lovetox: MUCs? I think Ge0rG knows a bit about that.

if i send a ping to the muc it answers with service unavailable, but im in it and it works

lovetox: you need to send a ping to your nickname

ahh

nice idea ^^

lovetox: and then wait for the ping response, not for a reflection of the request

lovetox: you need to ping periodically, and have a sufficiently high timeout (maybe 30-60s?)

hm why though? i have only the problem that gajim doesnt rejoin a channel if the stream management resume fails for some reason

so i intending only on reconnect to send a ping

lovetox: if resume fails, you can know for sure that you’re not in the muc anymore

Old yaxim used to ignore IQs when too many were received, which tripped on a series of self pings from poezio

(the server should’ve sent unavailable presence for you)

jonasw, yeah right ...

lovetox: what jonasw said

ok thanks for the help

if you *had* stream management before. If you didn’t, you cannot know really (unless you managed to replace your old resource, but this is trickerlyand).

*trickeryland

lets assume everyone has streammanagement :D

lovetox: you can't rely on unavailable presence, and some servers don't properly synchronize you if you send a join to an "already joined" MUC. Best workaround is to send presence unavailable right before the join

Ge0rG: you’re saying we can’t rely on "if the stream resumption fails, the server has sent unavailable presence and we can re-join"?

*re-join without trickery

And then you need to perform a magic mushroom rain dance and sacrifice a goat

or wait it out until MIX sweeps away all that tricky state :)

jonasw: I'm sure MIX will have its fair share of tricky state

Ge0rG: prevent it by reading the standard closely and fixing it before it becomes Draft :)

jonasw: let's see when I find another two hours

#iliketrains

I'm currently working on the Bookmark 2.0 XEP here https://lite5.framapad.org/p/OlHwr5GBlY (go to the bottom of the page)

"CAN" is not an RFC 2119 word

thanks :)

also, since you’re asking inline, MIX doesn’t need that anymore, MIXes are in the roster and annotated as MIXes

jonasw [20:08]: > Ge0rG: you’re saying we can’t rely on "if the stream resumption fails, the server has sent unavailable presence and we can re-join"? No, I'm saying that if you were in the MUC before, and aren't any more, it's better to send presence unavailable

jonasw, ok perfect

so maybe I should simply remove this part

Ge0rG: like, always, even if I left intentionally?

Ge0rG: why sending unavailable twice?

jonasw: no, in that case you'll be fine probably. Unless there was an s2s interruption while you tried to leave

edhelas: I am not deeply familiar with PEP and such, does PEP work well if different xeps building on it use pubsub collections?

Ge0rG: the s2s link hopefully uses SM as well.

well that where the definition of PEP is unclear

jonasw: I'm not aware of any s2s links with SM enabled

edhelas: at least, the node names are shared between different PEP usecases, so I’d worry about conflicts there

for me PEP can be a Pubsub node created under a user JID

jonasw, good remark

edhelas: another thing: I’m not sure if PEP allows multiple items per node. I think Zash mentioned that it’s not necessarily given that having multiple items is possible.

(well, it might *allow* it, but may not guarantee it)

Persisting multiple items, or persisting at all isn't guaranteed

jonasw, then 0277, 0330 and the proposal of OMEMO that I'm doing are broken :)

I don't remember if there were distinct features advertised for all types

edhelas: reference for roster-stuff: https://xmpp.org/extensions/xep-0369.html#mix-roster-capability-sharing

jonasw, yup found it, thanks

edhelas: have you considered <group/> for consistency with the roster itself instead of <tag/>?

jonasw, I do

(not saying that <group/> is a good name)

the thing is that I'm even wondering if this <tag> thing should maybe be moved to another XEP

why?

then we can actually tags Pubsub items

in general

hm

also i'd really like to be able to tag Pubsub nodes, is it possible to tag MIX as well ?

I sure hope that it’s possible to add <group/>s to MIX roster entries

we should ping steve about that

actually now I'm simply wondering if the Bookmarks 2.0 will not be useless with MIX

it may very well be

if it’s primary use case is MUCs

actually I was also dreaming about having browsers bookmarks sync over XMPP :D

edhelas: maybe ask steve about the <group/> thing for MIX roster entries on-list, i.e. if it’s possible to modify the groups of MIX roster entries like normal roster entries

I'll :)

thanks for all the tips !

you’re welcome!

I'll also continue to review the MIX XEP, I already made some comments and remarks on it

I'd really like to clarify everything to be sure that we cover all the use cases

yes

although at this point I think I need to implement it before I can further comment on it

personnally I don't see it directly as a MUC remplaceant

I'm planning to publish Atom items on it to really have a social network usage of it

I think it’ll serve as a muc replacement for many of the mundane uses

so now I'm wondering how we will be able to handle comments :D

or "threads" in MUC discussions

edhelas: re comments: https://xmpp.org/extensions/xep-0367.html

re threads: there is the <thread/> element from RFC 6121, but nobody has come up with a good UI for it yet

mhhh

(you could also use the <thread/> thing for comments in fact, if you control the participating clients and UI)

I think that this need to be clarified

I'll try to read on it

there is https://xmpp.org/extensions/xep-0201.html

are you planning to work on an implementation, in which client/server ?

I’m planning to work on MIX in aioxmpp

(client library)

ok

jonasw: not immediately necessary, but if you're going to make some changes to ecaps2 is there any chance I could get you to also add registry submissions to it? Eg. in the XEP that defines the registry it should have a little snippet of XML to include in the XEP (which is also what goes in the registry). Eg. https://xmpp.org/extensions/xep-0030.html#registrar-reg-features

SamWhited: ah, yes

If not I can make something up, but I figured you'd have more context and could write a better description.

thanks!

for some odd reason I haven’t seen those when I was looking at examples

btw, is anyone here using emacs to edit XEPs?

or spacemacs rather?

yah, they probably don't all have them

I need to update the disco-features registry to support provisional entries too; I just really hate looking at XML / XSLT so I haven't gotten around to trying to figure it out.

I like XSLT

let me know what you need.

is there a stream features registry?

ooh, if you really want too, the registries are located here: https://github.com/xsf/registrar

For the namespaces registry, I updated the vars.xsl and vars-xml.xsl file (which generate the HTML and the actual XML registry, respectively) to support a <status/> element on each entry, which would create a separate "Provisional" table for things that are still experimental (when things go to draft the provisional label goes away and they go into the actual registry)

the same needs to be done for features.xsl and features-xml.xsl

will do one I figured out how a stream feature registration shall look like

I'm not actually sure if there's a stream features registry

there is an XML file at least

oh yah, sure enough

I guess that one probably needs its XSLT files updated too

I also see from xep 53 that I should’ve appended :0 to the namespace

shall I do that now, while I’m at it, to have it consistent?

jonasw: You can if you want, but you could also wait until there was a breaking change and it would be fine; it's just a convention, but not having it (and then adding it) is the same as having it and bumping the version, it just changes the string :)

yeah

was just wondering.

jonasw: you like XML meta stuff? PARS still needs a schema, and I'd appreciate some support with that

Ge0rG: unfortunately, I’m not good at schemas

I only understand the very basics

SamWhited: added a commit to my PR, let me know if that’s the format that’s needed

Thanks! Works for me

SamWhited: I may be wrong, but I think Disco features already get a provisional table

jonasw: I don't see it, but maybe I forgot how this works

when I generate the HTML locally, I see this: https://sotecware.net/images/dont-puush-me/vZytdqh1MtbxSWpbTeSj9FZRULo_RCjCdudkVoN8rzY.png

oh, maybe I'm looking at the wrong xsl file

I think so. features.xsl is for stream features if I’m not mistaken.

yah, I'm looking at the stream features one

jonasw, I'm going to hijack your travis commit, k?

Guus: feel free

I totally forgot about it in the meantime

the build took more than a few seconds to start and then I forgot about it

it’s missing ; fi on those if lines

not any more...

ah cool :)

thanks!

don't thank me just yet...

ooh, green

but it didn’t run with two versions

so... if I now break one of the json files, it should fail again, right?

try replacing the whole matrix: block with: python: - "2.7" - "3.3"

isn't that what I did?

no, it’s: matrix: - python: 2.7 - python: 3.3 if I read the diff correctly

buidls with different python versions should look like this: https://travis-ci.org/horazont/aioxmpp/builds/213925500 (there should be sub-builds for each version)

aaah, yeah, sloppy copy/paste

https://travis-ci.org/xsf/xmpp.org/builds/214394838

that's more like it

linting now fails

is that to be expected?

ha right

that’s my fault

haven’t tested the linting with libraries

will fix, sec

harhar

simply proving that it works as intended. Good job.

you should claim that it was intentional.

right. totally intentional.

perhaps make the error messages a bit more verbose?

even more verbose?

so that random people understand what they need to fix

if they use update-entry.py, they won’t encounter that type of error

yeah, clear to people that are most likely to be unaware of the restrictions, but still need to fix the problem.

I'm assuming that most people will simply manually edit the json file instead of running that script. I know I will.

fair enough

what is the problem? those entries need to be in alphabetic order?

but, good job! this will be very helpful!

jonasw: feel free to squash some of those commits... :)

ugh

squashing will be interesting

meh, simply fixup of some of the odd travis commits

ah, cleaned it up a bit more thoroughly

SamWhited: https://github.com/xsf/registrar/pull/27

Guus: we need to talk what to do about those three or four pull requests for renewals which already came in in the mean time

I’m fine if you merge them, then I’ll rebase my updates to the tooling on top of them

and then we can merge the tooling updates. Even though I’d like to mention that this touches the template files and iteam may want to take a quick look at the diffs as due diligence

jonasw: that might be best

tobias, kev, intosi ^

even though the diff is very trivial this time: https://github.com/xsf/xmpp.org/pull/290/files#diff-01d6617e31aefee1df9589f998afb6de

I like how already, people are modifying their entries

indeed

lots of bigger and smaller website updates

heh, sco0ter links to http://xmpp.rocks which he redirects :)

let me know when you’re done merging

Done

are you addressing https://github.com/xsf/xmpp.org/pull/193 too, btw

(make the list sortable/filterable?)

I‘m not

could you? :)

haven’t looked into the magic done for the xep list

that’s verging into the space of javascript, which I try to avoid

jonasw: Awesome; thanks!

I know the feeling

Guus: updated my PR with new data

Guus: let us take that issue up when the renewal deadline is over.

then we’ll have a clear list of platforms

then we may decide how to present them

I have different ideas in mind, but they won’t work if we have a mess of platforms

whatever works

I don't care to much either way

heh, there's another client update :)

feel like rebasing once more, jonasw?

sure

although I‘d rather wait to a point where someone can merge the updates afterwards

?

well, it’s easier to rebase them in a large batch

less overhead

still not sure what you mean

but, if you rebase once more, I'll merge

ah!

then it’s fine :)

Guus: done!

I'm assuming you didn't sneak in anything nasty in this last commit... :)

you could grep over the touched files and see if anything outside data has been touched and take a quick look

git will even show it to you when you do the merge I think

travis passed

aw total missed chance to rootkit the server jonasw

jonasw, I clicked on your image earlier, do you run your own dnssec capable dns servers or what?

I rarely see the lock in my browser except on my own sites and debian.org...

moparisthebest: yes

although I wonder what image you’re talking about :)

jonasw, https://addons.mozilla.org/en-US/firefox/addon/dnssec-validator/

no DANE set up though :)

I guessed that this was the lock you were talking about, but not which image ;-)

oh it has DANE, but not for HTTPS

was more trouble than worth it

oh you rotate keys with letsencrypt too?

I keep the same key, just get new certs, and DANE pins my key

with the amount of HTTPS domains, it doesn’t scale to maintain the TLSAs. the gain is virtually zero compared to XMPP and SMTP (where I *do* maintain TLSA)

I don’t rotate keys, but still.

yea the gain is 0 because only the 4 people running that plugin validate them :)

but if you don't rotate keys it should be easy to add one record per domain

one day when I add DANE support to conversations it'll check it for https too

those are a lot of domains though.

if someone else could merge my renewals... (feels wrong to do it myself)

I can’t ;-)

jonasw, what server(s) are you using for dns? are they geographically seperated and such?

moparisthebest: pdns on a host in .de, I have backups in .at and in another facility in .de

allrighty

thanks for the had work jonasw

I'm off to watch some tv

you’re welcome

good luck

I'll need it

my wife promised me 'trash tv' :S

stay away from the news.

ohh, the trainwreck that's the US vote

that'd be good

jonasw, thanks I'm always curious about other's dnssec setups since it's such a pain :)

with powerdns it’s very trivial really. at least if you don’t mind keeping the keys online

it takes care of ZSK rollovers by itself

(and you can of course keep a backup KSK offline if you want to)

I run bind9 at my house as a hidden primary, then use https://freedns.afraid.org/secondary/, https://puck.nether.net/dns/dnsinfo, https://acc.rollernet.us/dns/secondary.php (2 servers) as 'free secondaries' to actually publically host it

could a couple of you +/- https://github.com/xsf/xmpp.org/issues/275 please?

so 4 different NS server locations, 3 different organizations, I feel fairly confident they won't all go down at once

I don't mind much not doing it either (although I think it'd be nice), but I'd like to know the general feeling

SamWhited: is https://github.com/xsf/xmpp.org/issues/279 this something that you can fix?

I feel too new to the whole thing to comment on that on the issue, but I don’t have any strong opinion one way or the other either.

moparisthebest: that sounds reasonable ;-)

jonasw, bind9 rolls over ZSKs by itself too, still have never rolled over my KSKs yet, but if it's good enough for IANA with .com and .org and such then it's good enough for me :)

Guus: I don't know how any of the website stuff is generated or where the refs come from, sorry

SamWhited Guus -- if those links are broken then we need to add them to the nginx redirect list to fix them

bear: yes please! i have more redirect candidates, but just went to bed.

k, let me see if I can get to that

💖

U+1F496 SPARKLING HEART

Unicode ...

Guus just add them to Issue #224

bear: Mix #224 https://github.com/xsf/xeps/pull/224

silly bunneh -- https://github.com/xsf/xmpp.org/issues/224

bear: Didn't consider multiple repos.

it's ok - I didn't realize it would auto-expand, now i'll be more careful

I can turn it off

nope - I don't think that's needed - again, it's just me not having been in channel recently to know better

IMO it's more helpful to default to XEPs than web stuff

The way it's written, the repo and the room JID are tied together. ( https://code.zash.se/riddim/file/tip/plugins/github.lua if anyone wanna poke around )