Your description makes it sound like there is a gigantic market where a commodity of ever increasing importance to customers is being massively underserved by market actors ? All that in the country where it is supposedly the easiest to incorporate and start competing ...

And I know that deploying fiber and stuff is capital intensive but considering the potential it still sounds like a huge opportunity .

I dont care all that much about the claims of the laptop being fully "libre" and I understand that a compromised microcode or FSP can compromise the entire PC, I do believe though that the less binary and closed components their are, the less attack vectors are available.
It may not be perfect but its better than every thing else which is available and it is a step in the right direction. Showing that some buyers DO care about such things demonstrates that a market exists for such machines even at a higher price.
If this can bring us good laptops which are not riddled by malware from day1 as most windows laptop are today, I say go for it.

Yes compromised microcode is still bad, puri.sm is supposedly working on a coreboot base for rev2 (they do acknowledge that rev1 shipped with an AMD Bios here : https://puri.sm/posts/librem-1...)
And yes they are still talking about coreboot instead of libreboot and I haven't had time to read up on what they are and why I should care. They also say that they are working with Intel to free the remaining bits which is laudable at least.

Regarding the memory, they have provided explanations regarding this exact point in their blog
"We heard from a backer that Intelligent Memory can run 32GB even though the specification states 16GB! This was corroborated by both PCWorld and our direct contact at Intelligent Memory this morning. We are back to offering 32GB, and will not change the existing orders down to 16GB."
See https://puri.sm/posts/32gb-is-... for a link to the pcworld bit

I am monitoring the librem 15 news very closely because on paper this is the laptop I want to buy except I haven't :)
All the reviews I have been able to find were done using prototypes, I have been unable to find any article on the production hardware for rev1, I have not seen anyone bragging that they had actually recieved their rev1 either. no pictures, no hands-on impressions etc no actual feedback on the build quality from non marketing sources.

While I don't mind the price for a really good quality laptop, I am unwilling to put it for a machine from a vendor with 0 track record on build quality, delivery, support, etc with no obvious return policy if the laptop doens't meet my standards
(a situation which is made even worse by my being european and puri.sm being an SF based startup which makes the chances of proper support / return pretty slim.)

struct reg_map {
  uint32_t reg1;
  uint8_t regbank[32];
  uint32_t filler[7];
  uint32_t reg2;
} __attribute__ ((packed));

struct reg_map *device = mmap(blabla);

device->reg1 = 0x1234UL;
uint32_t data = device->reg2;

You should try it some time :)

You're not being fair to Python.

Python's specific mantra (as listed in PEP 20) is: There should be one -- and preferably only one-- obvious way to do it. Emphasis mine.

Python, like any codebase, has a whole spectrum of "clean code" to "ugly hacks", but the richness of the language (and various libraries), provide a much richer foundation to let you avoid those ugly hacks.

(Unlike Perl, where the richness of the language somehow seems to encourage ugly hacks ^_-)

Actually, LEGO faced a decision whether they would go bankrupt or do tie-ins. The BI BI linked in another comment is excellent in showing what happened to Lego and their comeback.

All the crying about crappy tie-in Lego sets is hysterical hand-wringing. Yes, those occupy the majority of retail store shelf-space, but that only reflects the reteail store's decision. The key thing is that those tie-ins have not replaced other "pure" Lego sets in Lego's catalog. It's 2015, search online: there are many online shops and alternatives. Even better, there's Brickset, an amazing database of sets, which not only will show you the wide variety of still-in-production sets but also useful tools to help you find the cheapest set for cost-per-brick.

Rudy Rucker has some pretty crazy stories that always a blast to read (even though, or because you wonder what he was smoking when he wrote them).

One of those stories, Hormiga Canyon, has his protagonist build a computer cluster out of old cell phones, even using the phone's built-in voice recognition to control the cluster.

Does that count as Prior Art? :)

Actually, while they indeed compared two computers on the same LAN, they also included a computer on the internet. Furthermore, One of Dropbox's touted features is that it's able to detect and use peers on a LAN to avoid the unneecssary round trip through the cloud. I don't know about Google Drive, but judging by the results I suspect they can do the same.

And, more importantly, they compared the other clients on the same setup.

How you got modded "+4 insightful" is beyond me.

I used the SEO-deoptimized Awesome window manager. It is a tiling window-manager in the tradition of XMonad and Ion.

Since I started using it, I discovered how moronic is the concept of traditional window manager that allows overlapping windows: Either I want an app displayed, or I don't, and I certainly want an app to use as much screen space available automatically. I understand though that the traditional windowing approach is simpler to understand and see its point for the plebes ;)

I love its concept of "tags" instead of desktops, which gives me a powerful interface to mix and match which windows I want to display. I like that its "configuration" is actually a Lua program that allows me to precisely control how it behaves. I love that I can control it entirely through the keyboard. But I hate its stupid default keybindings: what's wrong with alt-tab nowadays?

Being a good teacher requires a particular set of skills I'm not sure Linus has, such as talking nice to idiots. (I kid! Treating (by definition ignorant) students like idiots is a fatal mistake). But seriously, being an expert in a field doesn't make you a good teacher, see: almost any college professor..

While this course will definitely get some name-brand recognition, I doubt it'll be better than a myriad other courses that exist already.

Portability also means giving up on system-specific optimizations and features. Some people have decided that Linux's market share means it's time to bank on those optimizations. Why not?

Indeed, but what information would you choose to store?

There's already a competing open standard.

It's what I use with my android devices (via BubbleUPNP), XBMC and my Squeezebox.

What was it?

It isn't useful on such a trivial example, but add in pointers...

int * func(char* a, char* b);
vs
int *
func (char *a,
      char *b);

(or better elaborate examples I can't be assed to come up with for a /. comment) ... and the milliseconds and frustration saved in parsing function declarations starts to add up