doctorsher

Here are the behaviors that I deal with on a regular basis:

(i) On mobile, I simply cannot view a native reddit video in landscape. When I turn my phone to landscape, it plays ~3 seconds of the video and then pauses. I hit the play button to continue, and it plays ~3 seconds of the video then pauses. Repeat ad infinitum. To clarify, this is not a buffering issue, as the video always plays when I hit the play button again.

(ii) On my mobile and laptop, their video buffering experiences holes in the buffer. It will be buffering, and after a certain point a gap is created between two buffered sections of video. When the video player reaches this gap, it is legitimately not buffered and stops playing.

(iii) Moving forward with the previous scenario, once it reaches this hole in the buffered video and stops playing, the playback stalls. I can sit there for minutes and it will not continue. I have to drag the slider backwards a few seconds, which will restart the playback/buffering process and then it works fine.

Years ago when I used this book, I was blown away by the clarity and concision of their explanations for non-trivial technical content. In particular, their treatment of scheduling stuck out to me. To my knowledge, no other OS textbook explains scheduling in such an intuitive way -- the typical treatment is more like "meh, here's the zoo of scheduling algorithms". In this book, they start with a long list of workload assumptions, and discuss the optimal scheduling policies under these assumptions. As they relax assumptions, the optimal scheduling policy changes.

Also, OSTEP seems to be in a better position to solicit feedback from readers / professors using the book than other operating systems textbooks. Remzi was quite prompt and appreciative when I sent some suggestions.

It sounds like a very important role of your coursework will be to get a more refined idea of what research you'd like to perform. I'd recommend taking classes in each of your areas of interest. The more advanced classes might give you a better idea of research in that area. These are probably the classes the PhD students will be taking.

I think a goal of 2-3 publications in one year is pretty lofty, but it depends how motivated you are and how much work you have to put into each of those publications. I think one solid publication (long paper, good conference/journal) will already give you a substantial edge in PhD admissions. This is a good goal; if you can do more, all the better! But I would hate for you to feel like you didn't meet your goal, and not be satisfied with one publication. Honestly, even if you go through the trouble of submitting a paper and it gets rejected, that is still an incredibly valuable learning experience.

Feel free to email / google hangouts me at adam.r.drescher (at) gmail (dot) com if you have more detailed questions, want my perspective during your graduate school experience, etc. I'm happy to help.

Do you already know your area of interest? This would be tremendously helpful for trying to pick an advisor.

Previous research experience might help, but there still are a number of factors that need to align in order to have a relatively smooth PhD. In my mind, it depends on the quality of your previous research experience, your advisor, and your goals upon graduation.

For my experience in particular, I did two summers of research at an R1 when I was in undergrad. Continuing on that path, I went to graduate school for my PhD. I am five years in; looking back, I was still extremely unprepared for the realities of research. When I did the undergraduate research, my hand was being held by the PhD students I was working under. Sure, I did some interesting stuff, but I received a substantial amount of guidance. Now in graduate school, my PhD advisor is very hands off, and our group tends to work in isolation. I have to push my research forward on my own. It took me years to get used to this, and not feel completely incompetent. My previous research experience was not a good predictor of my life as a PhD student.

In my mind, the best preparation would be to do a thesis-based masters degree and try to publish during that.

CPU / power usage is another big one. It's hard to justify a gain in throughput if it (i) reduces the application's CPU time, making it render the received data more slowly; or (ii) causes the phone battery to die faster because the network threads are constantly doing heavy lifting.

While I appreciate the size of the Ookla dataset, they need to be more thorough with their metrics. Phone performance is a multi-faceted problem, and throughput alone does not reflect the whole reality. Side note: dslreports has added bufferbloat testing a while ago, I can't believe Ookla still tests latency in isolation.

> someone who correctly points out that Google is not #1 in funding academic researchers

This type of wording is unnecessary. You are declaring that you are correct, while at the same time not providing supporting evidence when requested.

> This is precisely the sort of unproductive back and forth argument that the HN software is designed to discourage and punish

Welp. I am sorry this had such a negative effect on you. We can certainly agree to disagree. Cheers.

You are moving the goalposts. I am not saying Microsoft is not a prominent funder of CS research. I am challenging the notion they spend more than the NSF (or for an even stronger claim, the entire US Gov) on non-corporate CS research.

> But it's not surprising. There are all sorts of incentives for not breaking out those numbers. Still, the numbers are helpful. $14B is a lot of money. Even a small slice of that makes MSR one of the most prominent funders of CS research.

No, those numbers are not helpful. We have absolutely no idea what percent of that goes to academic CS research. It could be 10%, that would be incredible. It could be 1%, that is still great. It could be 0.01%, that isn't very notable.

> Funding grants for university faculty is NOT the only valid model for funding pure research. Therefore, we could very easily expect MSR to contribute a significant fraction of total US CS research funding without decreasing the competitiveness of the grant landscape for professors at universities. To me, the distinction between "basic research" and "everything else" seems more important than the particular funding model used. So the criticism that Google has a bad track record with long-running research agendas is highly relevant, whereas 'how much money goes to academics in universities' is far less relevant.

You are again moving the goalposts. The OP specifically said Microsoft was the largest funder of academic CS research.

I definitely agree. They have some top notch researchers, certainly.

I also agree that the budget comparison is apples and oranges, so I'm not quite sure why you brought it up. The OP made a grandiose claim about Microsoft's academic research funding, and made it quite clear with his discussion of Google that corporate research does not count. So I want a source for the amount of money Microsoft spends on pure academic research without the corporate ties.

Personally, I think if Microsoft spent anywhere close to a billion dollars on pure academic research, the funding situation in CS academia would be completely different. Microsoft grants would be pursued just as hard as NSF grants, because they would be such a major funding source. But that is not the case. Corporate grants are afterthoughts compared to the NSF for academics. Additionally, if Microsoft were plunking such a large amount of money into academic research, I assume they would at the very least publish that figure in a PR piece to cash in on some public good will.

To clarify, the whole networking stack isn't userspace. They still have the classical BSD networking stack. I think the userspace networking stack is used automatically by system libraries like URLSessions and this new Network.framework library. They introduced their userspace networking stack during last year's WWDC: https://developer.apple.com/videos/play/wwdc2017/707

WhatsApp tells their users and more importantly the US Gov they do not have any access to the message content itself. For this attack surface to be an issue, that would mean WhatsApp is telling a bold faced lie, secretly exfiltrating the plaintext to a 3rd party, and doing so without getting caught by any black-box auditing.

In the end, you'll have to do some introspection to see if the convenience gained is worth the privacy lost. How essential is it that you communicate in real-time with your relatives? Would your relationship be substantially different if you sent email or wrote letters?

From WhatsApp themselves [0]: "WhatsApp end-to-end encryption ensures only you and the person you're communicating with can read what's sent, and nobody in between, not even WhatsApp. Your messages are secured with locks, and only the recipient and you have the special keys needed to unlock and read your messages."

[0] https://faq.whatsapp.com/en/general/28030015