pyvpx

consistency and cleanliness. man pages are a dream, documentation solid and up to date. you can get things done and focus on what you want to accomplish, not what is getting in your way.

anyone not in SF/LA/NYC?

can anyone mention, recommend, or link to other layout algorithms? (preferably in the less than 1000 node/edge arena, but really any!)

it works, and the alternatives are far too costly from an opex/HR perspective.

can anyone recommend more papers/algorithms/reading material on layout algorithms in general? I'm very interested in layout algorithms of smaller node/edge counts (less than 1000). most material I've found is geared towards "big data" and hundreds of thousands (or more) of nodes/edges.

oddly, there is a growing trend of going userland and skipping native kernel because, well...the kernel is slow. userland can be made fast.

I think seeing Weave implemented in something like say, Snabb Switch, would at the very least be interesting -- from a performance standpoint.

I've generally only run with thinkpads, but I find OpenBSDs ACPI to be measurably better than everything else on power management.

you can still get hand made italian bicycle frames. you just have to know what you're looking for; it's actually not even that hard [0]

[0] http://www.pelizzoliworld.com/products/leggenda-for3/

I have no idea how Mechanical Turk works, however the end result in this situation would be the same. The independent contractor is allowed to choose the price by accepting tasks priced acceptably. the IC still "sets" the price they are willing to work for by only entering into agreements for work where they agree on the price being offered.

training and reconfiguration - and more pointedly B/OSS changes - are generally more expensive (opex) than upgrading equipment (capex)

I personally believe the biggest driver for IPv6 adoption will be Apple requiring support in iOS apps. Right now, IPv4 space is still comparatively "cheap"

having spent a considerable amount of time in the Horn of Africa, the reason African internet connectivity "sucks" is mostly obvious: lack of local talent, substantially increased costs, graft/corruption, and minimal use of peering infrastructure (plenty of peering fabrics, little to no engagement or use)

while I certainly love the excitement this has generated, it's not as easy as "we'll PoP Kenya! We'll PoP Tanzania! We'll PoP all of Africa! 50ms for everyone!"

you would think a place like Djibout with five submarine cables and terrestrial routes into Ethiopia, Sudan, and others, would have a vibrant peering fabric or CDN nodes.

Level3 just put a CDN node in there, and Telecom Italia Sparkle has been there for a while. AFAIK the CDN node is completely under-utilized and Sparkle provides high cost MPLS services for private and governmental outfits; not as much general IP transit.

More work needs to be put in getting carriers to peer at existing points of presence, and costs have to be lowered any way they can.

what makes you think it is trivial? how do I generate an accurate prefix list of who should be announcing what, when?

I was responding specifically to your "trivial to implement" comment.

I invite you read over the history of IRR and summaries of SO-BGP, S-BGP, PBGP, and MC-BGP.

that is because cloudflare has direct peering with many access networks. most providers see cloudflare routes directly and not through a transit such as Level3.

when the number of originated or announced routes are "reasonable", prefix lists, as they are called, are generated and applied to BGP sessions. when the originated or announced routes is substantial and/or fluid, there are little to no prefix lists; allowing anything to be announced and accepted.

in an ideal world that has not existed for over a decade, IRR data could be used to automagically generate accurate prefix lists but for all the usual reasons (inaccurate, stale data; obtuse interface; etc) that just isn't done on a wide scale.

You can put a prefix limit on a session (that's the "little" in little to no) but that doesn't stop someone from announcing 10,000 more specific routes to popular (YouTube, Facebook, Twitter, Akamai) destinations.

I've been buying OpenBSD CD's since 3.4 and a long time friend has commit bits with the FreeBSD project. I am a big BSD supporter :)

Bhyve is FreeBSDs...? I don't follow how it is misleading.

to the detriment of the citizens of Germany, as they pay more for less than just about anyone in the EU for mobile data.

considering the spectrum is a public resource, it's up to the government to make sure such a resource is adequately exploited for the good of the people. I think one could make a strong argument that extracting 1200EUR a head from telecom providers isn't the best way to exploit such a public resource.

Kabel Deutschland utilizes DS-Lite, effectively making the network IPv6 native. Comcast has been doing quite a bit of interesting work with going native IPv6 internally.

I am a huge proponent of LISP and this is one of the "coolest" uses of LISP I've seen yet: https://www.nanog.org/sites/default/files//meetings/NANOG64/...

such as?

the OpenBSD community gladly alienates those types of users. for "us" (I speak only for myself as a decade+ user and supporter), it's a qualitative game, not quantitative.

I think the email that was sent was incredibly rude. This person clearly spent considerable time writing it, and yet it shows from start to finish they didn't bother to read, research, or even understand the current state of OpenBSD.

Why should fools be suffered? OpenBSD developers have an OS to develop.

many OpenBSD developers use comic sans intentionally in their presentations.

if you watch closely (or really, at all...) the person serving you food doesn't touch the money. the tray is to take your payment and change between you and the central cashier.

care to suggest any resources that helped you reach wizard-like levels of usage? I very much wish to work towards that kind of ability in vi/vim.

in order to achieve 100Gbps at line rate, you have 6.7ns per frame. not a myth, but simple arithmetic.

take a look at network coding.

quite a few outfits do this with MPTCP, with varying degrees of success.