Conversations displays nested quotes down to the arbitrary depth of seven: https://github.com/iNPUTmice/Conversations/blob/467e34e2feb6a13c33ea24d9a1e9345689931a8c/src/main/java/eu/siacs/conversations/Config.java#L206
Anton L. Šijanechas left
Anton L. Šijanechas joined
Apollohas joined
alhas left
electrowindshas left
amee2khas joined
Beherithas joined
Beherithas left
Ingolfhas left
Samhas left
Samhas joined
dezanthas left
dezanthas joined
Wojtekhas joined
Anton L. Šijanechas left
jubalhhas left
Anton L. Šijanechas joined
wurstsalathas left
wurstsalathas joined
Stefanhas left
Stefanhas joined
amee2khas left
amee2khas joined
Anton L. Šijanechas left
Samhas left
flow
aren't limits like those mostly arbitrary? I mean, it's unlikely that someone did a survery and a histogram of messages to find a good cut of point
flow
otoh, if I think of mailing list discussions, a max quoting depth of seven appears a little bit to low
For instance, this image file is hosted at jabber.de. It can be shared from user(s) over IPFS instead of an HTTP server.✎
Schimon
For instance, this image file is hosted at jabber.de.
Instead, it can be shared from user(s) over IPFS instead of an HTTP server. ✏
Schimon
I think it __will__ be a definite feature that will qualify XMPP more than any of the centralized networks, even those that claim to provide "unlimited" storage, so called.
Apollohas left
amee2khas left
antranigvhas left
MattJ
Schimon, I think it then makes it impossible to expire/delete stuff? And in practice there is not much difference to today because people will still access the files through a small number of HTTP gateways
Schimon
Difference in what sense?
MattJ
To the centralization/decentralization aspect
MattJ
In reality it's still going to be served by jabber.de or an external IPFS host
MattJ
So what is it meant to achieve?
Schimon
> MattJ:
> 2022-06-15 06:32 (GMT+03:00)
> In reality it's still going to be served by jabber.de or an external IPFS host
Or both
amee2khas joined
Schimon
> MattJ:
> 2022-06-15 06:33 (GMT+03:00)
> So what is it meant to achieve?
Sharing large files.
MattJ
IPFS storage is unlimited?
Schimon
I didn't check. As far as I __see__ it, I thinl so, IPFS storage is unlimited, just like BitTorrent
Schimon
> MattJ:
> 2022-06-15 06:30 (GMT+03:00)
> Schimon, I think it then makes it impossible to expire/delete stuff?
Yes, it will be impossible to delete, unless everyone have deleted the file or disconnected all IPFS clients from the file itself.
Schimon
Essentially, IPFS is static BitTorrent. Please correct me if I'm wrong.
PapaTutuWawahas left
MattJ
It's similar, but your data is still hosted somewhere. In reality if you upload a large file, you're going to have to store it yourself. IPFS doesn't magically provide unlimited storage.
Of course. It's stored on users' machines. Each sharer is a host, just like BitTorrent (except webseeds which is http) ✏
MattJ
It offers an advantage if you want to publish something online and distribute it widely, publicly and ensure that content cannot be modified ever
Schimon
This too
MattJ
That's not usually the case for XMPP file transfers
MattJ
I'm not saying it wouldn't be nice to have a XEP that says how to identify an IPFS file (e.g. as an alternate URL)
Schimon
> MattJ:
> 2022-06-15 06:38 (GMT+03:00)
> So it offers no advantage for large files
Say, server limits upliad to 100MB.
Over 100MB, file is shared over IPFS
Schimon
> Me:
> 2022-06-15 06:41 (GMT+03:00)
> Say, server limits upliad to 100MB.
> Over 100MB, file is shared over IPFS
Schimon
This person attempted to retract a previous message, but it's unsupported by your client.
Schimon
This person attempted to retract a previous message, but it's unsupported by your client.
MattJ
But that 100MB file will still be uploaded to the same server, and downloaded from the same server
MattJ
99.9% of the time
Schimon
Oops. Sorry for the mess. Shall I repost my last message?
MattJ
I see all your messages, don't worry :)
Schimon
> MattJ:
> 2022-06-15 06:41 (GMT+03:00)
> But that 100MB file will still be uploaded to the same server, and downloaded from the same server
If user wants to share the 100MB file, it won't be possible to upload to server, so user's XMPP client will default to IPFS, no?✎
Schimon
> MattJ:
> 2022-06-15 06:41 (GMT+03:00)
> But that 100MB file will still be uploaded to the same server, and downloaded from the same server
If user wants to share the 100MB file, it won't be possible to upload to server (because server will reject it), so user's XMPP client will default to IPFS, no? ✏
MattJ
Using what IPFS server?
antranigvhas joined
Patigahas left
Schimon
> MattJ:
> 2022-06-15 06:40 (GMT+03:00)
> That's not usually the case for XMPP file transfers
We might have new use cases.
An entire community, exclusively using an XMPP groupchat, and IPFS to share contects.
I can see this japoen in Movim, or groupchat where there are a few participants that allow to post messages.
Schimon
> MattJ:
> 2022-06-15 06:44 (GMT+03:00)
> Using what IPFS server?
As far as I understand, every sharer is a server, no?✎
Schimon
> MattJ:
> 2022-06-15 06:44 (GMT+03:00)
> Using what IPFS server?
As far as I understand, every sharer is a server, no?
Sorry, I didn't look into the technicalities. ✏
MattJ
Okay, so you plan for e.g. mobile devices to participate in the IPFS network directly. That means if you share a 100MB file in a group with 100 people, you could end up sending 1GB data from your mobile device
Schimon
Yes. I see how problematic it can get
MattJ
It's possible it could be less, but very likely it won't be if every client requests it from your node
debaclehas left
Schimon
Sure
Schimon
But it's much, when bandwidth is limited, and I/O rate of SD card is higher...
Schimon
So, mobile clients will have to have a client-side mechanism for bandwidth-control
Schimon
> MattJ:
> 2022-06-15 06:40 (GMT+03:00)
> I'm not saying it wouldn't be nice to have a XEP that says how to identify an IPFS file (e.g. as an alternate URL)
It can be IPFS + HTTP.
Similarly to BitTorrent + WebSeeds (HTTP)
MattJ
Then you still have to find a HTTP server to host it in the first place :)
Schimon
Yes, if we choose so.
The goal is to reduce load from HTTP servers.✎
Schimon
Yes, if we choose so.
The main goal is to reduce load from HTTP servers. ✏
pep.
If you really want IPFS I can see XMPP servers proxying for clients maybe
pep.
But I'm not sure I see the point anyway
norayrhas left
MattJ
I feel like this is finding a solution before a problem. I don't think anyone is struggling with load on their HTTP servers right now. And I don't think many people are saying "I wish I couldn't delete my file uploads", and so on
norayrhas joined
Patigahas joined
Schimon
> pep.:
> 2022-06-15 06:53 (GMT+03:00)
> But I'm not sure I see the point anyway
Reduce load is the purpose of this
MattJ
The main thing that would be nice is to get some privacy, so that the upload server doesn't have to see every downloader's IP address
pep.
Schimon, Reduce load of what?
MattJ
But IPFS doesn't get that
pep.
MattJ, that can be done to a point with c2s server proxying right?
Schimon
Of HTTP upload bandwidth
MattJ
pep., yes, a simple authenticated HTTP proxy on the server (caching, potentially) would solve it
Ingolfhas joined
Schimon
> MattJ:
> 2022-06-15 06:54 (GMT+03:00)
> The main thing that would be nice is to get some privacy, so that the upload server doesn't have to see every downloader's IP address
This can be hapoen with IPFS, though peers can see IP, I guess
MattJ
Yes, peer-to-peer networks by their nature expose your network address to your peers
Schimon
> MattJ:
> 2022-06-15 06:53 (GMT+03:00)
> I feel like this is finding a solution before a problem. I don't think anyone is struggling with load on their HTTP servers right now. And I don't think many people are saying "I wish I couldn't delete my file uploads", and so on
Matt, I want to open an XMPP groupchat that only admins can post and share contents of 2GB and above.
I think IPFS uploads are a great solution for that
MattJ
First you need to find a host (IPFS or not) which allows 2GB uploads
Schimon
I'm the initial host
MattJ
or if you're running IPFS on your own device, a good internet connection
MattJ
There's no reason you can't do this already today, there are HTTP<->IPFS gateways, and you can just share a URL
MattJ
and then serve from IPFS on your local machine
Schimon
Or, I can re-share IPFS link (then again, in that case I can just send the link)
Schimon
Ok, I'm beginning to sound silly. I'lm take a break and think this through✎
Schimon
Ok, I'm beginning to sound silly. I'll take a break and think this through ✏
MattJ
I think being able to specify multiple URLs for the same content in XMPP would be a nice feature
Schimon
Ditto
MattJ
That way you could share HTTP and IPFS, and clients that understand IPFS could use that
Schimon
Of course. That too.
I still do sense that incorporating IPFS will cause higher interest in forming new communities inside XMPP.
nikhas left
Kev
Doesn't SIMS allow multiple references?
xnamedhas left
Kev
Maybe I misremember.
emushas left
emushas joined
debaclehas joined
Schimon
SIMS is an XEP?
Anton L. Šijanechas left
MattJ
This about summarizes my feelings about the distributed web hype:
MattJ
4 years ago, someone posted their blog to HN, about how their blog was hosted on IPFS: https://news.ycombinator.com/item?id=18027958
MattJ
It's no longer loading, tested via multiple IPFS gateways
MattJ
But guess what? The wayback machine has it covered: https://web.archive.org/web/20180919220141/https://ipfs.io/ipns/Qme48wyZ7LaF9gC5693DZyJBtehgaFhaKycESroemD5fNX/post/putting_this_blog_on_ipfs/
Schimon
Haha, wayback usinh an http<->ipfs proxy
Schimon
> MattJ:
> 2022-06-15 07:22 (GMT+03:00)
> It's no longer loading, tested via multiple IPFS gateways
Perhaps if he was just as motivated you are (xmpp community members), it would still hold.
MattJ
So from this small random sample, if you want some data to be persisted for years: it's better to submit it to wayback than host it on IPFS
Schimon
MattJ: you Kev pep. Link etc. It looks like you're inside XMPP for life ;)
Patigahas left
MattJ
I'm not sure what this has to do with XMPP :)
Schimon
Just kidding a little to raise up the fact that you are consistent, in presence
cyrilhas joined
Schimon
> MattJ:
> 2022-06-15 07:25 (GMT+03:00)
> So from this small random sample, if you want some data to be persisted for years: it's better to submit it to wayback than host it on IPFS
I still think IPFS a worthy addition
Schimon
I've over 600 incomplete torrents.
Some last, some not.