To take a slightly cynical view, but a view I honestly believe in, on the want to switch to JSON: XML looks old, most programmers have shallow opinions and chase new things.
It's not parsing performance, computers are plenty fast for text IM. It's not bandwidth, the difference between JSON and XML is negligible after compression. It's not developer ergonomics, any sane programmer is using a library that abstracts either format. It's not compatibility with the domain, as XML wins for XMPP due to namespaces.
The answer is that XML looks old. New programmers (half of all programmers have less than 5 years of experience) grew up in a world of JavaScript where XML was "legacy" since the day they arrived. In reality it's kept working fine, while the volume of software has increased around it naturally using the trendy tools. They've not looked back to understand why it was made or why it has merits, it's already got negative connotations and they're caught up in the new stuff. The new stuff has merit too, that is why a programmer is wise to respect both.
Also, what use is a stable and mature XML ecosystem when you can earn big nerd points by reinventing XPath for JSON the 5th time?
XML and JSON only started about 5 years apart (1996 vs. 2001). I think XML's adoption was more broad more quickly, so the gap felt a bit longer than that (XML was a core part of my jobs from 1998-2007 and JSON ever since). I think more of what makes XML look so much older is the fact that so many new projects have been picking JSON over XML for 10-20 years now, and that's why younger folks have seen much more JSON than XML.
My software engineering career started long before either, and I have used both extensively. While I don't think that it makes sense for most well established projects to switch, just as rewrites often don't make sense, I would almost always pick JSON for new projects. Your points are valid, but one part of the ergonomics that you didn't mention is that it's so much easier for humans to read and write JSON, thereby speeding up development, debugging, testing, etc.
Is it really a question about processing performance though? I’ve always assumed it was about bandwidth and latency… Communication protocols don’t need to be human-readable because tooling has always provided that at a higher level. A binary protocol just as a text protocol like S-expression or the divine simplicity of IRC is just as digestible when documented. And there are better facilities to have extensibility regardless. I think we can all agree it’s as much a failure as a success if we’re still talking the same points 25 years later.
I agree that it should not use JSON. I think it would be better to use DER (or SDSER if streaming is needed) rather than XML or JSON, but that is just my opinion. However, if the existing protocol already uses XML, then that is what it will be, but the new one perhaps would not use XML.
Here's a thought - the thing that will ultimately doom XML, and perhaps JSON as well, will be the extra expense XML incurs when being tokenized for LLMs.
If they did adopt JSON then it wouldn't be XMPP anymore.
As a user, I don't care much. But my experience with XMPP is that was not as solid as other solutions, including closed source ones. I could've been issues in clients' implementation, but overall it wasn't great
> my experience with XMPP is that was not as solid as other solutions, including closed source ones.
Considering that WhatsApp is essentially an old/non-federating fork of ejabberd (from where this blog post originates), I think XMPP is doing alright in that regard.
Readit News
It's not parsing performance, computers are plenty fast for text IM. It's not bandwidth, the difference between JSON and XML is negligible after compression. It's not developer ergonomics, any sane programmer is using a library that abstracts either format. It's not compatibility with the domain, as XML wins for XMPP due to namespaces.
The answer is that XML looks old. New programmers (half of all programmers have less than 5 years of experience) grew up in a world of JavaScript where XML was "legacy" since the day they arrived. In reality it's kept working fine, while the volume of software has increased around it naturally using the trendy tools. They've not looked back to understand why it was made or why it has merits, it's already got negative connotations and they're caught up in the new stuff. The new stuff has merit too, that is why a programmer is wise to respect both.
Also, what use is a stable and mature XML ecosystem when you can earn big nerd points by reinventing XPath for JSON the 5th time?
My software engineering career started long before either, and I have used both extensively. While I don't think that it makes sense for most well established projects to switch, just as rewrites often don't make sense, I would almost always pick JSON for new projects. Your points are valid, but one part of the ergonomics that you didn't mention is that it's so much easier for humans to read and write JSON, thereby speeding up development, debugging, testing, etc.
As a user, I don't care much. But my experience with XMPP is that was not as solid as other solutions, including closed source ones. I could've been issues in clients' implementation, but overall it wasn't great
Considering that WhatsApp is essentially an old/non-federating fork of ejabberd (from where this blog post originates), I think XMPP is doing alright in that regard.