jdev - 2019-09-19

Now how do I display this DOAP thing properly?

Link Mauve ^

Ge0rG, what is your wanted rendering thingy?

Something like that? https://pulkomandy.github.io/xmpp-doap/samples/movim.xml

Link Mauve: yes

Something like that? https://linkmauve.fr/extensions/xep-0048.xml

Link Mauve: I've forked your xmpp-doap repo and included the style, but it doesn't show the table content

Ge0rG, then just use the linked XSLT style sheet, by viewing source.

Ge0rG, this was PulkoMandy’s.

Link Mauve: I can't do that on a local file :(

Maybe it isn’t fully up to date with the latest format?

Ge0rG, oh?

yeah, local files are a separate security domain

Indeed, it isn’t up to date.

I can fix that.

Link Mauve: can we also get the XEP titles in there?

Ideally, I'd also just have `<XEP-0077/>` instead of `<xmpp:xep rdf:resource="https://xmpp.org/extensions/xep-0077.html" /><xmpp:xep rdf:resource="https://xmpp.org/extensions/xep-0077.html" />`

Ge0rG, the XEPs don’t have any semantic document thingy we could use to extract the names from.

Also <XEP-0077/> isn’t valid RDF.

Link Mauve: there used to be an XML file listing all that.

Link Mauve: that was a placeholder for whatever the shortest possible syntax is

No idea how I ended up with XMPP

Ge0rG, syntax doesn’t matter here.

Semantic web parsers already exist, and reusing one is much less work than writing a parser from scratch.

I had a much shorter syntax at first, but it wouldn’t be compatible with existing DOAP software.

So syntax *does* matter? ;)

anyway, this is just minor nitpicking.

Ge0rG, I mean, shorter syntax doesn’t matter.

I want to get this thing going, and that means I need to render it locally

Ultimately you can have something generating your doap file :p

Link Mauve: what would you need to allow XEP name resolution?

pep.: I had some*body* generate it.

Ge0rG: good first step

Ge0rG, you can use xsltproc from libxslt to generate a HTML file from the CLI.

Ge0rG, hmm, a deeper understanding of how XSLT works.

Also since there is a Python version nowadays (see https://github.com/xsf/xmpp.org/pull/594) it could be done much easier.

Link Mauve: xsltproc shows the same output as chrome. Your style must be b0rken

I thought you had the bestest changes, but now...

It’s not mine, complain to PulkoMandy.

> Resource interpreted as Stylesheet but transferred with MIME type application/xslt+xml: "file:///usr/src/yaxim/xmpp-doap/style.xsl". Is that bad?

Maybe my XML is b0rked? https://op-co.de/tmp/yaxim.doap.xml

ah, yes it is.

> The requested URL /style.xsl was not found on this server.

Zash: it's all horrible!

XSLT is pretty cool

but it was <implements> instead of <xmpp:supports>

https://xmpp.org/extensions/xep-0199.xml

Link Mauve: is <xmpp:version> the XEP version, and <xmpp:since> the application version?

This is super-confusing.

Yeah that's the semantics

Suggestions?

Ge0rG, https://github.com/pulkomandy/xmpp-doap/pull/6

Ge0rG, yes, version and since have this semantic, as you can see in the OWL schema.

Ge0rG, https://linkmauve.fr/ns/xmpp-doap

what do i need to get started with doap? is there a good sample file including style file and everything?

Daniel, easiest would be to start from one of the ones linked from here: https://github.com/xsf/xmpp.org/pull/594

Link Mauve https://hg.prosody.im/trunk/raw-file/tip/doc/doap.xml

Link Mauve, but that doesn’t do the rendering

Board meeting with doap stuff is today right

Daniel, rendering would be up to consumer services, such as xmpp.org, JabberFR’s wiki, some random comparison table, etc.

right. i'm just looking for something that makes writing easier

Zash, hmm, the format changed to better match DOAP since then.

and verify that i did everything correctly

Zash, see https://lab.louiz.org/poezio/poezio/commit/31c605528c0ad8b22d617073fb2e1722b82fabb2

Meh

Daniel, clone https://github.com/pulkomandy/xmpp-doap/pull/6, then run `cd samples; xsltproc -o conversations.html ../style.xsl conversations.xml`.

That’s the closest we have from a renderer.

Zash, when did you do that, and from which example?

Link Mauve, thanks. that probably comes close to what i was looking for

If you additionally want RDF validation, https://github.com/xsf/xmpp.org/pull/594 contains a doap.py script which you can use to parse it using a real RDF parser.

https://github.com/xsf/xmpp.org/pull/594/files#diff-64332f4e20f843debc79815b0ecd6a97R81

Link Mauve some week(s) ago. Dunno what it's based on, could be the translation thing I made for the old xmpp-features thing.

Link Mauve, is there any chance we can get XEP revision into there?

Wasn't there already?

XEPs change, sometimes substantially

MattJ, it already is there.

Oh

It’s the version property.

For instance here: https://pulkomandy.github.io/xmpp-doap/samples/movim.xml

The second column.

Oh yeah, it's just the Dino one I looked at didn't have it

Link Mauve https://code.zash.se/xmpp-features/file/6198147cec64/doap.lua

Zash, right, that was for an older version, which wasn’t really good design wrt RDF.

😕

The main difference is that there is no wrapper Software anymore, instead it’s the Project which implements a SupportedXep.

Before the Project softwares a Client, which doesn’t make any sense in either English or RDF.

How2fix?

Do you want a patch or guidelines?

Either works.

So, remove the <software><Client/></software> wrapper, namespace-aware-sed s/xmpp:supports/doap:implements/ and that should be it.

Also add a <category rdf:resource="https://linkmauve.fr/ns/xmpp-doap#category-client"/> category.

Link Mauve, how can i find a list of available categories?

Daniel, I have a short list at https://linkmauve.fr/ns/xmpp-doap, some other people have been using the GNOME list.

I’m not fully sure how that’s meant to work.

Link Mauve, pep.: can we rename <xmpp:version> into <xmpp:xep-revision>?

because "since" and "version" are utterly confusing.

Ge0rG, that’s pretty redundant though.

Link Mauve: not for a human looking at the XML

The SupportedXep is version ~something~.

Link Mauve: not for a human looking at the XML

Ge0rG, a human should be looking at the RDF, not at the XML.

That is, at triples rather than at elements and attributes.

Link Mauve: right, because "Version" and "Since" in the table header is NOT AMBIGUOUS

Ge0rG, you can rename the table header though.

It was a simple example PulkoMandy wrote to verify the format.

Link Mauve: okay, so let's assume for the moment that it's implemented-xep-version. Now tell me how implemented-xep-since makes any sense.

Ge0rG, can you suggest improvements as patches on top of https://linkmauve.fr/ns/xmpp-doap?

I’d also like this to be hosted by the XSF eventually.

So all files will have to be updated anyway.

Link Mauve: as I said, I'd do s/version/xep-revision/. But you seem very much opposed to it, so I don't think a PR will help here.

I’m not very much opposed to it, I’m just not sure I see the point.

it makes editing and understanding easier.

And I’d like to avoid unnecessary breaking changes until it gets moved to xmpp.org.

less context you need to keep in RAM

feel free to postpone this change to when it moves.

so what's a good file name for that? doap.rdf?

That’s a good one yeah.

That way web browsers will serve it as application/xrd+xml, which is ideal.

application/xml should work too.

Link Mauve, https://raw.githubusercontent.com/siacs/Conversations/master/doap.rdf

Daniel, maybe remove the comments affecting poezio.

Hum, like Zash you based it on an older example. :/

that's the one you linked me to

poezio’s doap file is https://lab.louiz.org/poezio/poezio/blob/master/data/doap.xml

So hmm, as I told Zash at 13:57:10, remove the <software><Client/></software> wrapper, change xmpp:supports into doap:implements, and that should be it.

So I named it .rdf, and now Chrome tells me to "Save as" the file:// URL

Link Mauve: I've done that, and I'm using your update branch and it doesn't render the table

Ge0rG, can you show me your DOAP file?

ah, I took `doap:` too literally. `<implements>` did the trick

Link Mauve: how do we handle Compliance Suites?

Good question.

Maybe the same way Zash’s XMPP-Features did, by checking for complete status on the relevant XEPs?

Or for now we could say complete and mark the profile(s) in the <note/> field.

Link Mauve: what about a separate RDF element?

Ge0rG, that could work, do you have something in mind?

Link Mauve: no, I was just asking

i think it should be figured out automatically

to me that's the point of having machine readable files

Daniel, yes, indeed.

But which entity would be responsible of that?

Daniel: I agree, but it's probably more work to do the automatic figuring-out

Yes but only once

What if the developer has a different opinion?

Ge0rG, https://code.zash.se/xmpp-features/file/tip/compliance.lua

These rules are what XMPP-Features was doing.

Link Mauve, does that look ok now? https://github.com/siacs/Conversations/blob/master/doap.rdf

Daniel, yes, the structure looks good. :)

Maybe make 0166 partial?

why?

You say it’s only for file transfer.

But anyway it’s your call.

sure but 0166 doesn’t mandate you to support a specfic application, no?

Looks great, I’ll add this URL to my xmpp.org PR.

how can i get and post requests to xmpp with postman

postman is the wrong tool for the job, even though in theory you could do some weird BOSH stuff

really

i just want to be able to monitor http requests

xmpp is not http

i know

but it uses http though

like 5443

I don't know what you're talking about now

i want to be able to listen to requests on a specific port

and i was told postman was the best program to use

It doesn't use HTTP, and what port it uses is not directly related to what protocol it uses

You can of course tunnel anything over everything, hence BOSH as Ge0rG mentioned.

You can do XMPP over HTTP, but most XMPP connections are not doing this (unless they are from a web browser)

You can do HTTP over XMPP as well.

Is there a solution yet for the F5 on Converse when running via WS?

Ge0rG: 198 I guess?

Zash: we had a discussion about 0198 on anon, which doesn't work

It works fine if the server assigns the same JID to the same user

which can be done through evil means, such as cookies

you can have cookies on WS?

Sure

The initial handshake is a HTTP request, you can pass whatever headers or parameters there.

That's evil indeed, but I'm only slightly surprised.

MattJ: do you have such a mechanism in prosody yet?

I'm using BOSH now for the webchat because of this.

Ge0rG, not open-source, sadly

I think there is no real downside of keep using bosh for those anon cases

Daniel: you are probably right

DONRASH111