jdev - 2020-03-27

To check supported xeps on a specific server one needs to be logged in, correct?

Jeybe, depends, simply querying the disco#info of the server's address may already reveal most XEPs

but obviously not the ones that are only exposed to users of this service

To mention, I'm completely new to XMPP development (and so the conrecte XMPP and XEP specs) and mostly new to development in general. The XEPs I'd like to check are the ones from the Compliance Suite

Jeybe, well you could simply get the source code of the compliance checker and look how the checker does it

Jeybe: some relevant features like Push and MAM are probably only visible on your own account JID, not on the server JID

so yes, you need to have an account

Ok thx

Makes sense, as for the tester on compliance.conversations.im also needs account credentials

the code behind is https://github.com/iNPUTmice/caas

thats why I am in favor of stuffing those features also in the disco#info response of the service's address: you don't need to have an account to check how widespread an xep/feature is implemented

left: `"<root xmlns=\"root_ns\" a=\"b\" xml:lang=\"en\">meow<child c=\"d\"/><child xmlns:ns0=\"child_ns\" d=\"e\" xml:lang=\"fr\"/>nya</root>"` right: `"<root xmlns=\"root_ns\" a=\"b\" xml:lang=\"en\">meow<child c=\"d\"/><child xmlns=\"child_ns\" d=\"e\" xml:lang=\"fr\"/>nya</root>"`'

People doing xml properly in their stack, what version would use the less bytes in the general case?

Here for sure it's the right one, but in XMPP it's quite common to reuse jabber:client right

So I'd rather not shadow it

to

many

quotes

yeah sorry, that's the output of my tests :)

pep this is about XML serialization, right?

yes

I think I would simply use the prefix if it already exists, otherwise not

left: `#"<root xmlns="root_ns" a="b" xml:lang="en">meow<child c="d"/><child xmlns:ns0="child_ns" d="e" xml:lang="fr"/>nya</root>"#` right: `#"<root xmlns="root_ns" a="b" xml:lang="en">meow<child c="d"/><child xmlns="child_ns" d="e" xml:lang="fr"/>nya</root>"#`'

here, without the escaping

flow, not what I'm asking

or maybe with this you mean I shouldn't shadow

and prefer the left version

where is something shadowed in the left/right examples you showed?

In the right example the child xmlns shadows root's

the second child

Which means if this child has descendants, I won't be able to reuse root's ns without redeclaring it

right

but in xmpp we tend to not re-use jabber:client once an extension element namespace has been declared

<forwarded/> being one prominent exception

An exception would be forwarded or MIX.

ahh MIX, really?

MIX-PRESENCE IIRC.

cause its in pubsub potentially

anyway, I am not sure what you are asking, or if there is a generic answer to your question

Presence in PubSub! \o/ <items node='urn:xmpp:mix:nodes:presence'> <item id='123456#coven@mix.shakespeare.example/UUID-x4r/2491'> <presence xmlns='jabber:client'> <mix xmlns='urn:xmpp:presence:0'> <jid>hecate@shakespeare.example/UUID-x4r/2491</jid> <nick>thirdwitch</jid> </mix> <show>dnd</show> <status>Making a Brew</status> </presence> </item> </items>

but if your codebase is so far feature complete that you think about saving bytes… ;)

flow, no I'm just curious, I have to make a choice :P

pep., what I’d do would be to let the user declare the prefixes they want on the root <stream:stream>, if XEP-0044 negociated it’s allowed, and only use prefixed elements if their namespace has been defined there.

My implementation does left at the moment

But our tests do right

pep: smack would (potentially) prefix presence in this case

with the rule I mentioned before

let me test on real data

Link Mauve, minidom doesn't have the knowledge of "stream header" or not

pep., yeah, so it should be handled at another layer, which is why I said “let the user”.

Sure, with the new API I let the user define custom prefixes

<stream:stream xmlns:stream="http://etherx.jabber.org/streams" xmlns="jabber:client"><message><body/><replace xmlns:ns0="urn:xmpp:message-correct:0"/></message></stream:stream>

hmm, ns0 is indeed useless here.

And message and body just reuse the default ns

wait, this is not valid. it should be ns0:replace