And for a sample of the writing, here's the section on compiling the CLI tool jq to WebAssembly: http://levelupwasm.com/sample-jq.pdf (though note that it builds on previous material so it doesn't re-explain everything)
This provides access to hundreds of public and tens of thousands of restricted access cancer genomes (and often the matching "normal" samples): so you can see and visually verify mutations. Interestingly, too, you can check out germline polymorphisms which might predispose someone to get cancer.
Basically all the work is done in the browser : gzip decompression, custom parsing of sequencing alignment, png generation, etc. Lots of data are pushed to the client at load time like gene location information for genome build "hg38" (the latest). Server only provides reference sequence and sample genomic alignment "slices" for the region of interest. So, all the state is handled in the browser, server only provides trivial "wget" requests.
It is mostly wasm (source is C compiled with emcc). The front page is the requisite html/javascript start page. Pages are created in wasm code and pushed to JS for updating the DOM. Lots of calls to emscripten_run_script("your javascript here"). DOM manipulation straight from Rust/Go/C would be cool but it is not here yet.
Upside is server simplicity and security ... and no downloading and setting up complicated software.
There's no reason they can't. Public/private partnerships are done all the time.
In Chicago, AT&T got a deal to put small cell towers in places they wouldn't ordinarily be allowed in exchange for building the Office of Emergency Management and Communications a private wireless data network.
In the subways, there was some kind of deal struck with the wireless companies, as well. I don't remember exactly how it worked but it was something like Brand X gets 2-year exclusive on wireless service in the subways if it builds the infrastructure that everyone else can piggyback on later.
It was slightly comical years later when a dozen cities around the world were touting how awesome it was that they were "first" to have wireless service in their subways, while people in Chicago had already been living with the scourge of loud people Facetiming other nobodies about absolutely nothing for everyone to hear. The screeching of the L wheels is preferable.
Air drones - loud, need to be pretty big and noisy to lift larger payloads eg. groceries. In order to cover use cases of food/grocery/package delivery across a city and surrounding suburbs there would be a constant stream of high powered drones whirring through major areas and around streets/footpaths. I don't think this would be acceptable to the public at scale.
Self-driving footpath size delivery vehicles - people won't tolerate anything going faster than 10km/hr on sidewalks and that + crossing roads makes it too slow. In addition, at any reasonable volume and covering longer distances eg. 5km trips across a city these would end up blocking footpaths eg. if grocery delivery sized drones were working in any sort of numbers.
Self-driving delivery cars - could work, but to replace all current package/grocery/food deliveries would involve a huge fleet of cars, or carrying multiple deliveries per car which makes them laggy - not good for takeaway food or packages where a person needs to go to a sidewalk to meet the delivery vehicle.
The idea is that delivery drones in tubes can travel at high speeds - faster than cars (50+km/h with few/no stops), can be centrally controlled to cooperate well on limited bandwidth routes, and will end up being able to be routed like internet traffic. being able to arrive "just in time" will make them ideal for sidewalk pickup by the end customer.
Although highly capital intensive, I think that a well-planned network like this would allow efficiency and competition the likes of which have never been seen before in delivery of groceries, packages and takeaways. Imagine the ability for ANYONE in the city surrounds to compete on grocery prices, not just the large supermarkets on high value real estate. "Last mile" delivery costs are more than half of total courier costs - imagine if these were reduced to near zero.
This means on average overall each person gets 1 delivery per day. Assuming delivery cost of $2 per delivery, this means in a city of 1 million people there is $2 million daily revenue for operators of this, or $700 million per year. At a cost of capital of 5%, this would allow up to $14 billion dollars to construct this network in a city of 1 million people. At an average $4 delivery cost there would be up to $28 billion available.
What percentage of the city would have easy access to tunnel endpoints? It's a pretty difficult last-mile problem.
At least in the short term, a good bet would be funneling stock to retail stores. In lots of neighborhoods, big trucks parked on narrow streets to offload cargo is a big problem.
To do this you would need buy in from all the legacy carriers first. It doesn't work without having them all as your partners, so you get the economy of scale you need. That will be very difficult. Next, you need to consider your own operating costs. How are these vehicles going to be powered? Routed? You will still need to sort and distribute the packages. What if one breaks down in route? How are you going to lift them back up onto the surface? Are you planning to retrofit buildings with tubes going down into your network? Or distribute in/out terminals throughout the city? Both which would require an insane amount of upfront politicking and planning. There will be issues of easements, zoning, and liability to work out. Then, even if you get all the players to agree to your plan, which is unlikely, you still need to deliver the package at a (significantly) lower cost structure than they currently are able to do with government subsidized roads.
Why underground? Could it work in above ground pipes? How big would be the packages and robotic vehicles?
Packages, grocery, and food bring different requirements in terms of delivery speed and size, which could lead to different solutions--some easier to address than others. What is the pain point and most desperate customers for each?
I agree, tunnels below ground have the disadvantages of:
- expensive to bore out
- cannot be changed
- expensive to maintain
Above ground pipes would allow for easier maintenance and a much cheaper roll-out. Crazy but could be worth looking into existing infrastructure like power cable routes etc..
It's not just an issue of speed. The new gmail design is very hard to read/understand. Red-on-pink is not the best contrast choice. The icons feel fuzzy without the button borders, are hard to decode, and are too tight together.
Strangely, while the icons are all packed tightly, the email list wastes extra space vertically, and requires extra scrolling. Not the best design choice.
The high-contrast theme brings gmail close to the classic design, but it does not work well enough to make the icons more recognizable and recover the wasted space in the email list.
A key reason to own a Mac would be to build iOS apps and MacOS software.
For me, the reason to own a Mac (besides hardware support and generally liking macOS) is the amount and quality of third-party software. There is a lot of software that I either need or want, and there are no good-enough competitors on Linux, e.g.: Microsoft Office (for work), 1Password, Affinity Designer, Pixelmator, Acorn, LaunchBar, Little Snitch, Tweetbot, Mic Snitch, DeckSet, Paprika (for recipes), Things, PDF Expert, Arq, Dash, Sonos, etc. In addition, a lot of these tools and macOS are very well integrated, e.g. I can directly search Dash or 1Password from Launchbar. I can make a phone call to a person directly from Launchbar, etc.
For some work there are better open source tools, e.g. Emacs as an editor, org-mode for notes/outlining, LaTeX for typesetting, Handbrake for video transcoding. But they all work fine on macOS.
Oh, that same old argument. No they're not. Absolutely not even close. This is coming from someone who actually uses linux 90% of the time every day. Linux on the desktop is still a huge mess, every desktop environment on linux is a real pain to use, I won't go into details, but those who use linux daily on the desktop cringe reading comments like this, unless using a wm like dwm or i3, but those suck BIG TIME too if you ever have to leave the terminal.
Or you know, to have a UNIX, but also a nice just-works-most-of-the-time desktop UI, device support (for stuff like external audio interfaces, scanners, etc) and access to all kinds of proprietary pro apps, from Above and Microsoft to whatever.
For the State of Hawaii, incorporation fee is $50 for for-profits, with a $15 annual business report fee thereafter. The annual report is a 1-page form stating total shares issues, officers, type of business, and addresses.
For entrepreneurship, I would recommend: (a) Technology Entrepreneurship, and (b) Venture Deals with Brad Feld -- both are team and project-based experiential courses offered via novoed.
Technology Entrepreneurship is focused on the lean startup. The next session is expected around June.
Webassembly for genomics sounds cool :)