keithwinstein

14 ms·

61.
by keithwinstein 5y ago
I think it mostly has to do with: 1) the large "implementation surface" of three components (TLS+HTTP+WebSocket), each of which by itself requires an API more complicated, if provided by a userspace library, than a kernel-provided
62.
by keithwinstein 5y ago
That's the current salary for a Stanford CS PhD candidate who stays for the summer in a "90%" research assistantship (as is typical in some subdisciplines), but (a) obviously the cost of living here is high, and I think some
63.
by keithwinstein 5y ago
Couple of suggestions: - A lot of people have written really good advice about the application process; see https://blog.keithw.org/2016/02/stock-advice-for-undergradua... for a compilation. - In the U.S., univers
64.
by keithwinstein 5y ago
Thanks! The basic video-over-WebSocket technique was part of our paper here: https://puffer.stanford.edu/static/puffer/documents/puffer-p... Talk here: https://www.youtube.com/watch?v=63aECX2M
65.
by keithwinstein 5y ago
Hmm, we're getting <200 ms glass-to-glass latency by streaming H.264/MP4 video over a WebSocket/TLS/TCP to MSE in the browser (no WebRTC involved). Of course browser support for this is not universal. The trick, which
66.
by keithwinstein 5y ago
No, a compromised intermediary can't fool Guardian Agent with false values for (a), (b), or (c). The client checks/enforces these. The details of how it works are in the linked paper.
67.
by keithwinstein 5y ago
ObPlug for Guardian Agent, which is basically "safe" ssh-agent forwarding (and works with Mosh and SSH): https://github.com/StanfordSNR/guardian-agent The basic story is that ssh-agent really just exposes a p
68.
by keithwinstein 5y ago
Yes, PulseAudio and JACK do this as well -- there is a DBus API that makes them give up an ALSA device to an application that wants direct access. http://git.0pointer.net/reserve.git/tree/reserve.txt
69.
by keithwinstein 5y ago
I'd love to hear tptacek's view on the pros and cons of "authenticated connections," where the client uses asymmetric crypto to establish an authenticated-as-the-principal bytestream, and then issues a bunch of unsigned
70.
by keithwinstein 5y ago
I don't think this is quite right. My understanding is that 4:2:2 chroma subsampling for Rec. 601 video (and the choice of sampling at 4x the chroma sub-carrier) came around in the days of 525/59.94 vs. 625/50 systems, befo
71.
by keithwinstein 5y ago
Poynton has a pretty plausible-sounding explanation here ( https://poynton.ca/PDFs/Chroma_subsampling_notation.pdf ): "The commonly used leading digit of 4 is a historical reference to a sample rate roughly four ti
72.
by keithwinstein 5y ago
We have the "append mode" -- run `mosh --no-init` and it does what you'd like.
73.
by keithwinstein 5y ago
It's tempting, and thank you for the PR, but, do we then have to push a new Mosh release every time there's a Unicode update? And the users have to get this release installed on both the server and their client? It seems... like i
74.
by keithwinstein 5y ago
How much do you want? We'd very much welcome your involvement -- we'd love to have you start helping out in the IRC channel, help us put together a credible plan for the next release (probably helping the current maintainers), and
75.
by keithwinstein 5y ago
Re: Unicode, please see https://news.ycombinator.com/item?id=28151652 . (Mosh really just uses your C library for knowing the width of characters -- if the server's and client's libc support a character, Mosh will
76.
by keithwinstein 5y ago
Please see https://news.ycombinator.com/item?id=28151652 , but happy to discuss further if you want.
77.
by keithwinstein 5y ago
Unicode (really, knowing the width and printability of any given character) is mostly up to the C library on the client and server -- Mosh has historically stayed out of that. Unfortunately Apple is not always super-great about keeping thei
78.
by keithwinstein 5y ago
I think the truth is that we don't feel like we need a release. We've never had a security hole, our track record speaks for itself, and I wouldn't really want us to rush anything and blow that. There are some features that
79.
by keithwinstein 5y ago
Mosh original author here -- fun to see us back here yet again. Happy to answer any questions.
80.
by keithwinstein 5y ago
Heya, great to see you pop up here! I gotta be honest -- I think EC2 (and, in general, doing computation in units of VMware/Xen-style virtual PCs) is the actual hipster compute substrate. AWS Lambda feels closer to cgi-bin from 1995, i
81.
by keithwinstein 5y ago
Thank you for this nice writeup! This paper was led by my student Sadjad Fouladi ( https://sadjad.org/ ), part of a broader theme of coercing a "purely functional-ish" design onto everyday applications. There's
82.
by keithwinstein 5y ago
I'm not an expert in UMTS-style cellular networks, but I think the numbers in this article are misguided or incorrect. :-( E.g., the article seems to conflate "Internet-style" measurements of practical round-trip latency over
83.
by keithwinstein 5y ago
Looks reasonable! On a system with virtual memory, one popular trick is the "virtual ring buffer," which lets the reader and writer always access a contiguous region of the full requested length. The idea is to map the same backin
84.
by keithwinstein 5y ago
This list is also a bit incomplete, or more likely just refers to classes taught in the current school year. I co-teach a different ethics-focused class in the Stanford CS department (CS 181, which in this incarnation started in 2017). I th
85.
by keithwinstein 5y ago
Steve was the kind of local institution that makes MIT such a special place to be around -- a nice and incredibly knowledgeable guy who ran the swapfest, attended the radio club meetings, brought his generator to field day every year, and l
86.
by keithwinstein 6y ago
I knew Matt Yourst (author of PTLsim) when we were housemates at MIT; he died tragically a few years ago. I'm really happy to see that his software lives on.
87.
by keithwinstein 6y ago
Heya, we (Mosh) are fans of Browsh and are happy to talk if we can be helpful! It's definitely doable to expose (and directly link with) the Mosh networking and terminal libraries if you're unhappy depending on the executables. Yo
88.
by keithwinstein 6y ago
One more small piece of advice: try to get on whatever mailing lists you need to to start attending the research talks at your college or university. (Some of this might be harder in the age of Zoom/economic collapse, but still...) Thi
89.
by keithwinstein 6y ago
My advice: don't worry too much about picking a field (and definitely don't focus too much about what people who happen to be the same age as you are up to -- it's a big world). 'elonmollusc's advice is really sound
90.
by keithwinstein 6y ago
I saw this movie last August at a showing at the Computer History Museum and one thing really confused me. The documentary includes a dramatized 1980s interview between a fictional interviewer ("Michelle", played by Kaliswa Brewst

More ›