text
stringlengths
0
1.32k
**Mikeal Rogers:** Right. So I've spent some surprising amount of time looking at this space, because it's one of the few areas that we haven't seen a huge amount of growth in.
**Adam Stacoviak:** \[40:03\] Haven't seen what?
**Mikeal Rogers:** A huge of amount of growth in, the kind of growth that we see in other places. I've spent a lot of time actually figuring out what we could do here and what our current barriers are. One barrier which I think is a problem and it needs to get resolved, but it's not the number one problem, but it is a ...
This is kind of hilarious - now that Brandon has a company that's doing Crypto in the browser, he is shifting some of his focus... Because he does a bunch of standards work as well, and there are some new 64-bit stuff and some new math stuff coming out that Brandon is pushing forward into TC39; also, the Node Foundatio...
There's one right now that obviously I couldn't be at, but I went to the last one that they did and resolved a lot of the module stuff, a lot of the concerns that we've had about how do we integrate these module systems; we were able to work that out. And I'll continue to go, to start to try to resolve some of those pr...
Another thing we have to think of is should we try to reach a system like R and Python have, or should we try to go about this in the way that has brought us a lot of other successes? If you look at the early days of Node, we weren't able to bind to a lot of low-level C libraries, and because we weren't able to do that...
Mikola Lysenko, who's here, he was a math PhD and got really involved in the Node.js community and started building different algorithms as just single modules. So you have this modular way of looking at things, which is not the way that we've done math and science computing before. There are these giant libraries in C...
So do we wanna enable a broader ecosystem of people building that kind of stuff out? I think we currently have the most compelling ecosystem for 3D computing, I think, all built for WebGL, obviously... But Mikola has a project called Regl, which is a phenomenal suite of toolchains for doing modularized 3D programming. ...
I don't think that we're going to figure out a way to have a better Fortran binding layer, or cross-compile these Fortran libraries. R is mostly Fortran code. It's a layer of bindings on top of a lot of very old Fortran code that SciPy also binds to and their libraries bind to. We're not gonna get to a point where that...
\[43:58\] I think lastly SciPy - R as well, but SciPy, I really look at... They did a ton of outreach and community building with the academic community to get them onto SciPy. Over the years, I've met a few of the people that have really pioneered that work. It's hard community work, you're working with a lot of older...
Python is not uniquely good at scientific computing; I mean, not a knock against Python, it's just not better than any other particular language for that. They don't have great bindings to all these old Fortran libraries, it's relatively slow... They've just built a great community and a great support system for it. Th...
**Adam Stacoviak:** How does that actually play out though, at an institutional level? How do you play out that support? Is it similar to the Apache Foundation where they kind of adopt projects and they incubate them to a degree and then take them on full-term once they matured to a degree? Is that some sort of plan? H...
**Mikeal Rogers:** I don't think that we have a lot to offer at the project level, mainly because there are actually a couple non-profits now that have spun up to help back a lot of this work; SciPy isn't one of them. Those institutions are set up to take in grant funding as well, because a lot of the ways that those l...
**Adam Stacoviak:** So if someone's listening to this, that's like...
**Mikeal Rogers:** If you're at a university, If you're at an academic community, definitely reach out to me and Tracy. I'm @mikeal on Twitter. Also, the foundation bylaws... I think almost all of the Linux foundations that have been set up under the LF have an allotment for a membership tier that can be for non-profit...
I think a big part of this is that you have really passionate people at different academic institutions that love Node.js, that love the web, that push for these technologies. What we don't have really is a venue for them to talk with each other about what's working and what's not. I think that they probably have more ...
**Adam Stacoviak:** So we're two years past the biggest shift in the Node community, which was the fork of io, merger back into Node, Node Foundation, several iterations of this conference, year-over-year growth at a hundred percent rate, so a growing community... You've touched on educational pieces, you've touched on...
**Mikeal Rogers:** \[48:05\] That's a lot of hot points, isn't it?
**Adam Stacoviak:** Yeah, that's a lot of highlights, I think, and it's pretty close, but just in case there was something that we couldn't leave this conversation without you talking about...
**Mikeal Rogers:** One thing, if we're looking at the future of Node -- I mean, if Rod were here, he could talk a lot more to the technical side of things, but I'm definitely more focused on the institutional level work that we've been doing. We've done a lot to re-message and promote Node. By that, I mean a lot of the...
We've really built that out since the foundation started. Because we've taken this broader message and we've really focused on the connections, we haven't been as aggressive in framing how good Node is at a few particular things... Serverless is obviously a huge growth area for us. If we did nothing, it would probably ...
I think that in the future you're gonna see us talk about these verticals a little bit more and be a little bit more assertive about how good we think that we are in those particular cases.
**Adam Stacoviak:** Well, Mikeal, that's all I've got, man.
**Mikeal Rogers:** Yeah, great.
**Adam Stacoviak:** It's been great. The future of Node is bright... I'm glad to have you in the position you're in; I know it's been a long journey getting here, but fun times ahead. Great community, great work.
**Mikeal Rogers:** Thanks!
• State of HTTP/2 implementation in Node
• Impact of HTTP/2 on Node's performance profile and APIs
• Decision on whether to include HTTP/2 in NodeCore or as a module
• Definition of web fundamentals and criteria for inclusion in Core
• Importance of keeping NodeCore small and focused on basic web protocol support
• New features and complexities introduced by HTTP/2, such as stateful header compression and multiplexing
• Discussion around Node.js and HTTP/2 protocol
• Importance of security in Node.js with HTTP/2
• How HTTP/2 supports better security through compliance to specification
• Browsers requiring TLS for HTTP/2 connections
• Limitations of Node's current TLS performance
• Excitement about developers using HTTP/2 and its potential
• New features and applications enabled by HTTP/2, such as push streams
• Criticisms of the HTTP/2 protocol, including staple header compression limitations
• Binary coding vs stateless compression for headers
• Complexity of HTTP/2 protocol (flow control, prioritization)
• Server affinity issue with HTTP/2 and proxy software vendors
• Performance benefits of HTTP/2 (efficient socket use, reduced failures)
• Tradeoffs between HTTP/2 features (e.g. header compression, TLS) and API complexity
• Breaking changes to Node's API for HTTP/2 adoption
• The challenges of adopting HTTP/2 as it requires deliberate design and implementation.
• Potential use cases for HTTP/2 include server-to-server communication within protected environments.
• James Snell is working on the HTTP/2 implementation, but it's open-source and contributed by multiple developers.
• The Node organization is developing HTTP/2 under a GitHub repo (github.com/nodejs/http2).
• It may be possible to use both HTTP/1 and HTTP/2 simultaneously, with clients negotiating which protocol to use per socket.
• Deprecating HTTP/1 is considered premature due to its fundamental role in the Node ecosystem.
• Implementing HTTP/2 as a native module is being explored as an alternative to integrating it into Core.
• Request for help from the community
• James Snell mentions stabilizing and testing aspects of a project
• Areas where help is specifically needed (tests, performance benchmarks)
• Repository link provided for reference
• Conversation importance due to Node Interactive's closure
**Adam Stacoviak:** \[01:05\] In this episode I talked with James Snell from IBM, the Technical Lead for Node. James is also a member of Node's Technical Steering Committee, as well as the Core Technical Committee. He is currently working on Node's implementation of HTTP/2. I talked with James about the state of HTTP/2...
\* \* \*
**Adam Stacoviak:** So what's the state of HTTP/2 in Node? I know you're working on it now, you've recently tweeted about a prototype server...
**James Snell:** The current state is just trying to figure out how it would work in Node. There's a lot of new things within HTTP/2, it's a brand new protocol, even though it's got the HTTP semantics, request/response, headers and that kind of thing, on the wire it's very different, so it requires a completely new imp...
**Adam Stacoviak:** We had a discussion earlier with Thomas Watson and Sam Roberts from IBM... Sam was really passionate about talking about keeping Node small, and Thomas actually coined - I don't know if it's him or not - the term SmallCore. One of the discussions we had in that conversation was what should or should...
**James Snell:** Personally, I think it should be in Core. The reason for that - Node has always been a platform for web development; there's always been that web server.
**Adam Stacoviak:** That's true.
**James Snell:** It's a primary use case, even though there's so many different places Node is being used, and in different use cases, a lot of it always goes back to having Node. If you look, there is no standard library in Node, but there's HTTP, there's URL parsing, there's support for these fundamental web protocol...
Now, if HTTP/1 wasn't already there, I wouldn't be thinking that we should add HTTP/2.
**Adam Stacoviak:** You'd think module at that point.