421 karma · joined March 1, 2015
We have a 3.0/0.5 DSL connection which can support 1 high quality or 2 medium to low quality Netflix streams. Web browsing is tolerable when everyone is online, but the second the uplink bandwidth is saturated (0.5Mbit/s) everything else grinds to a halt. I think you'd notice it too if your connection weren't symmetric. That is, your connection was 50/5 or 50/10.
We were on-track to get a 50/50 fiber connection (the service supports up to 1000/1000) - however the ISP hasn't yet started to build the network out to our village - a year later than they said it would be available. In contrast, our neighboring villages all have their service, but there is a risk they will pull out since the build cost is 3x what they forecast (narrow country roads where the only option is to dig up the road, rather than run cable under verges or sidewalks).
FWIW, the build cost is a measly £600,000 - or $929,670, which in my estimate works out at between £600 and £1200 per resident.
Whilst there is nothing I particularly dislike about Spotify, I found it's 'Offline Mode' particularly clunky.
When I play a song on Apple Music that I have in a playlist or album, I can replay tracks without the need for a persistent 3G/4G/WiFi connection. Whilst you can do this in Spotify, I understand you have to manually click a button to enable offline playback.
Furthermore, Apple Music will shows me which tracks are cached for offline playback, since those which are not are greyed out and un-clickable, whereas I don't recall seeing anything like it in Spotify. This feature is particularly useful on the tube (London Underground for people in the USA/Europe/Rest of World) where you have intermittent WiFi at stations (in fact, by the time your phone has connected to the access point on the platform the train is already pulling out of the station - so you've got to be fast!).
However, I feel like Apple could have done better with the Apple Music experience on iPhone and iPad. For example, what is the 'For You' tab all about? Just some bouncing spheres with labels of different genres? I quite like the 'New' tab since I can quickly find popular music, but it feels too much like the iTunes Store, where you click-to-buy.
Yet, I really like the artwork they've set up for artists, and I think the player is excellent. I can't say I use 'Connect' that much, but it is an incredibly convenient way to follow artists you like and, if you're lucky, hear some B-sides that didn't make it onto the album.
I haven't yet decided whether I'm going to continue my trial with Apple Music or go back to Spotify. I just want a solid music streaming service that will 1. let me listen to the artists I love and 2. help me find other artists who I will love. I feel like neither of the two fully satisfy those requirements just yet.
Edit: I forgot to talk about Beats 1. I'm really enjoying listening to Zane Low on Beats 1 radio after his move from Radio 1.
> FYI, this happens because SSH automatically presents a public key to the server when trying to authenticate. If the server doesn't know that key, then SSH tries the next one. You can enumerate all of someone's keys this way (like this SSH server does)
Therefore, even though I can't authenticate as Linus Torvalds, an SSH server can see me present his public key and hence, log that public key for future use, like sending a message? Is that not correct?
Thanks for the authoritative reply!
If Flash doesn't run under NaCl why use the PPAPI? Wouldn't libc be much more convenient, since you don't have to port Flash to the PPAPI?
Furthermore, I would imagine using the PPAPI directly wouldn't do much to enforce security since Flash can do whatever it likes in it's virtual address space, including jumping past checks and assert instructions in the PPAPI?
If Flash doesn't run under NaCl, does it run under the same user and privilege as Chrome?
Finally, can you comment on this quote from http://krebsonsecurity.com/2015/07/adobe-to-patch-hacking-te...:
> Several reports on Twitter suggested the exploit could be used to bypass Google Chrome‘s protective “sandbox” technology
It's frustrating that there is no source. However, the above quote would suggest that Flash runs inside NaCl. Is this factually incorrect?
I recall that Chris Rohlf presented an implementation defect in one of the two at a Black Hat conference some years back, but the sandboxing is otherwise solid.
Let's say I downloaded every public key from GitHub to ~/.ssh. Would this identify me as everyone on GitHub, or just the owner of the first public key to match?
Furthermore, I wonder how Go channels compare to libevent (more specifically, epoll/queue) for high-performance network software. Is there any previous work which compares the two?
The reason Native Client exists extends well beyond Adobe Flash Player. Native Client exists to enable your browser to run native code from any website (think compiled C and C++ code) without it being able to compromise the security of your machine.
All interaction between a Native Client module, the host process, and the host the operating system must be mediated via the PPAPI (Pepper API). Native Client has a set of rules which are enforced via the validator to ensure that untrusted code (that is Adobe Flash itself) cannot jump, load or store outside to addresses outside the sandbox. More information on how Native Client does this can be found in the links below.
Note: you should be familiar with the x86 and x86-64 instruction set for it to make much sense.
Nevertheless, here is some more information:
[1] https://support.google.com/chrome/answer/108086?hl=en-GB
[2] http://www.eecs.harvard.edu/~greg/cs255sp2004/wahbe93efficie...
[3] http://static.googleusercontent.com/media/research.google.co...
[4] http://static.googleusercontent.com/media/research.google.co...
First, as 20 something I can't imagine how hard it must be to wake up with the realization that you no longer have a job. Second, I think you deserve some real credit for 1. identifying that you were unemployable and 2. fixing it.
However, the real trick is identifying when you're unemployable (or even starting to get close to it) before you're unemployed. This is a game - and one that you'll never really win. But like all games: you keep playing - you get better. The trouble lies in that this is a game in which you need to be ahead of the gamemaster (that is - current and future employers).
> I do think if I found myself unemployable again I could repeat...
I want to stop you right there. I think your objective from here out is to make sure you never find yourself unemployable again - rather than waiting and/or hoping it doesn't happen.
> almost any kind of expertise in a major JavaScript framework like Angular, Ember, or React/Flux might get you a remote job fairly easily
These are valuable skills and expertise to have - and they will land you a job - but don't let it make you a one tricky pony - as that's a fast-track ticket to unemployment. I think few of these front-end frameworks will be here for the long-haul.
If I were in your shoes (of course I'm not in your shoes - it's 100% your choice) I would take what you've developed so far and start to dive into the fundamentals and then get the expertise.
For example: get to know the architectures of these frameworks: analyze and criticize them, attempt to improve them if you can; learn design patterns and identify the anti-patterns (important !); learn your algorithms - as it is these features that make you a valuable professional in the long run. Not just another PHP/JavaScript/ColdFusion dev.
Then you can wade in deeper and build up deep full-stack expertise. You could think about:
"Oh how can I optimize view rendering in Angular?" "How can I solve head-of-line blocking when using web sockets?" "What is the optimal compression algorithm in order to serve JavaScript files for mobile devices?"
I can't imagine a huge number of people think about how to solve these challenges - but I would bet that a lot of people face them.
All the other stuff is just tools to enable you to apply your expertise easier and faster.
What will happen is that two TCPs will share the available bandwidth (or close to it) via AIMD. However, if you've ever uploaded a file on an asynchronous Internet connection (i.e. a DSL connection), saturation of the uplink kills effective throughput for other TCPs (I presume because ACKs queue up at the router and are either delayed or timeout, resulting in fast retransmit if you're lucky, and a timeout at the sender if you're not).
> "does not slow down your internet connection" as it uses a "limited portion" of idle upload bandwidth."
How do Microsoft know what your upload bandwidth is? When do Microsoft know when that upload bandwidth is no longer idle?
Is there an RTT estimator that stops the upload if the RTT exceeds a threshold? How would Microsoft know what the RTT should be for that customers network link?
Perhaps they can detect consecutive timeouts of TCP segments? That would be interesting because the TCP I know provide the abstraction of a reliable, in order transmission over an unreliable network - so why would this API call exist? Perhaps it does exist? Can anyone comment?
I think this raises more questions that it answers?
Furthmore, why is it when you delete a photo from your iPhone it can persist on iCloud for up to 30 days?
Likewise, why is it when I restore my iPhone and set it up from scratch it starts to download 100MB + of data over WiFi? [1]
[1]. When I setup my iPhone from scratch (Apple ID, iCloud disabled), it hammered my WiFi at the full RX rate of my DSL line for approx. 10 minutes. That's 4Mbps * 60 * 10 = 2400Mbits = 300MB. I made sure that my iPhone was the only device connected to my WiFi. What could be in that 300MB of data? I should proxy all traffic from my WiFi to my DSL modem via another PC and do tcpdump on that!
The latter can be issued via the OEM or carrier - whereas the former is issued by Google. Since Google own the trademark on the Android name (http://developer.android.com/legal.html), perhaps Google can enforce a rule that an OEM which ships it's device without compliance with the above cannot call their device an "Android device".
$page = $_GET['page'];
include ($page.".php");
Whilst allow_url_include (http://php.net/manual/en/filesystem.configuration.php#ini.al...) was set to false, I could still craft a URL like:http://example.com/?page=admin/index
which expanded to http://example.com/index.php?page=admin/index where the real admin was at http://example.com/admin/index.php and offered complete access to the backend without authentication or authorization - let alone other files in the file system.
In another project, I found that the server had register globals turned on, and therefore could craft a URL like:
http://example.com/admin?valid_user=1, where valid_user was a PHP variable set to true iff their session cookie could be authenticated in the database.
I think it's terrifying that these things still make it through to production websites
Could (and should) a safer language pre-empt that you should first check strlen(str) != 0 before doing strcmp? I know of no language which does this.
> Input validation can be made to be machine-verified
Likewise, can you elaborate on this? How does the machine know what type of input this is? If input is tagged by the developer, what's to stop the developer from forgetting to tag the input correctly?
If so, have you ever forgot to use the virtual machine and instead browsed or read email on your host operating system? If so, what did you do?
Furthmore, is it possible to break out of the hypervisor and into the host operating system?
Sorry, are you referring to Native Client or Chrome's site isolation?
1. Networks and protocols
I know most of the TCP/IP stack, physical through to application layer, but lack expertise in most of it. For example, it couldn't tell you how to pack a query in a DNS packet even thought I know the header and body format - or tell you how TCP Fast Open works, despite knowledge of the 3 way handshake.
Security and exploits:
I'm really interested in exploiting software. In fact, just today I developed a buffer overflow vulnerability in a HTTP server that can be exploited by a maliciously crafted HTTP request (a dummy server for educational use)
Operating Systems:
Likewise, I have working knowledge of operating systems but nothing exemplary. I could talk to you about interrupts, ISRs, user level v.s kernel level threads, the VFS, etc..., but lack hands-on experience with any of them.
Finally, and to a lesser extent (I haven't got that far yet) the Linux kernel.
It's a lot, I know. However, it's becomes obtainable when you realize people dedicate years to their craft. You just have to be persistent, self-motivated and enthusiastic - the rest takes care of itself
Despite my indifference to Node.js, I can see the unification of client-side and server-side scripting into a single language being incredibly valuable to the millions of PHP and JavaScript developers (both front and back-end) out there: most of whom beyond the realm of HackerNews and want to get it done.
Not only that, it would significantly reduce the barrier to entry for web scripting and disincentivize beginners from picking up all those nasty habits such as md5 password hashing and manually constructing SQL queries from raw input which are so worryingly accessible.
The only thing I can see preventing this are:
1. Lack of a mod_node for Apache. Perhaps this exists? 2. The upload and refresh workflow so many PHP developers are familiar with 3. A scheduler to prevent any one website from blocking the Node.js event loop on shared hosting
Is this completely bonkers, or is it quite sane?
[1]: https://en.wikipedia.org/wiki/Aero_Spacelines_Super_Guppy#/m...
I did find the same presentation but from another event: https://www.youtube.com/watch?v=d6pU4C4PVoY
However, I cannot recall whether this was a subject of the talk or whether I asked it in Q&A. I think it was the latter.