I have experience with all three, but I disagree that you can take away a lot from them after Linux. An OS from a user perspective is essentially a package, configuration and service manager. And to learn something new and valuable in either of those categories you'd need to go into very different direction. Like docker, nix, guix for package and configuration manager, erlang supervision trees for services.
90 hours/week = 18 hours/day (assuming 5-day work week). That leaves only 6 hours for other chores, travel and sleep. This is not sustainable.
So I'll assume people who can pull off a 90-hour work week work all 7 days of the week which would amount to 13 hours/day. That still leaves sufficient time for a 7-hour sleep and 4 hours of chores and entertainment.
Now, if you are your own boss where you decide the scope and the deadlines, the work is going to be fun and I can understand that you would not mind working 13 hours/day. But I don't think working 13 hours/day for a boss or a company that considers me a cost in a business transaction of buying my skills is going to be fun.
To provide an anecdote for the effect this has on your life. I worked 70+ hours minimum (106 was my "record") for 5 years and averaged 4 hours and 45 minutes of sleep per night. It was spread out over 6 days a week which made it a little easier.. but it wasn't worth it.
I couldn't keep any intimate relationship, pretty much just dated very casually over that time.
Rarely saw my family.
Saw my friends so infrequently, that I may as well have lived 2,000 miles away. It was a big event any time I saw them which was usually once every 2-3 months.
My blood pressure reached full blown hypertension levels and I needed medication.
It's not a healthy way to live and I've undoubtedly erased years from the end of my life as a result. I don't care how motivated or passionate you are about what you do, this work load isn't sustainable and will forever be a "Dark Ages" era of my own life.
I'd need to be convinced that they had a really interesting problem, one that would keep me awake at night. I'd also need to be convinced they understand the idea that "anything we can solve by throwing money at the problem isn't a real problem". Real problems can't be solved by money and problems that can be solved by money aren't real.
You'll find at a job that a 70% productive day is outstanding. Time not lost to meetings, travel, etc. is rare. Working from home helps because it extends the work time and minimizes the interruptions. My years working from home were the most productive.
As for the 70 hours thing, I work 7 days a week. Since I have no outside constraints the term 'day' gets a bit blurry; sleep happens when it happens, eating happens between keystrokes, etc. In fact, I'm "on vacation" this week at the Outer Banks but reading a PhD thesis. My wife can't understand why I brought "work" to a beach resort. She works in medicine so this really is "escape time from work" for her.
Interesting work is its own motivation. We seem to have lost that idea somewhere. Her work is not interesting so she needs to escape. Mine is interesting and I consider not working as "lost time".
Andrew has the possibility of "interesting work" but I'm going to bet they screw it up with things like "required sensitivity training", meetings, "required plans and schedules", "interviewing new employees", "friday lunch gatherings", skip-level interviews, open-plan floor office space, investor presentations, morning standups, on-call phone coverage, skype meetings, slack interruptions, agile card games, progress reports, company security protocols, random phone calls, trip reports, trip refund paperwork, employee ratings, changing goals, offsite meetings, time tracking, parking issues, pair programming, etc. All of which are examples of time-sinks I've experienced on the job. I can't spend 3 weeks absorbing a relevant PhD thesis "between meetings". I can't write really difficult programs without huge blocks of uninterrupted time.
Andrew Ng mentions in the job posting that they want dirty work as well as design work to be done.
"Since we're an early stage company, you should be flexible in your tasks and do whatever is needed, ranging from dirty work like data cleaning, to high-level work like algorithm design."
Can you hire temp work to do data cleaning? (Aka, solve it with money? with mechanical turk?). If so, don't waste my time. If the data cleaning requires judgement then that's fine. Every job has dog work. I have to chase references for the papers I read which takes days of effort. But doing it right (aka "providing quality") is important here. Quality matters.
p and t are implicitly defined cardinalities (i.e., the smallest cardinalities with given properties). See Definition 1.1 of the source article (https://arxiv.org/pdf/1208.5424.pdf)
> Briefly, p is the minimum size of a collection of infinite sets of the natural numbers that have a “strong finite intersection property” and no “pseudointersection,” which means the subsets overlap each other in a particular way; t is called the “tower number” and is the minimum size of a collection of subsets of the natural numbers that is ordered in a way called “reverse almost inclusion” and has no pseudointersection.
In short, there are two infinities, p and t, that are implicitly defined by some characteristics. It was previously known that there was a relationship between them, but it was not suspected that there were, in fact, equal.
Considerable work is required to understand what these are, and what the result means.
And to answer your explicit question, yes, the reals can be put in 1-1 correspondence with 2^N, the power set of the naturals.
> In a breakthrough that disproves decades of conventional wisdom, two mathematicians have shown that two different variants of infinity are actually the same size
I thought there are only two types of infinity and Cantor already proved that they are different.
* Uncountable infinity which is the cardinality of the set of real numbers
* Countable infinity which is the cardinality of the set of integers
Cantor has already proved that uncountable infinity is larger than countable infinity.
Is this article claiming that mathematicians have proved that these two infinities are equal now? Doesn't that contradict Cantor's proof? What's going on here? Is the Cantor's proof flawed or have we introduced a contradiction to mathematics?
If you take the power set of an infinite set, you /always/ get a cardinality bigger than the original set. So there are an infinite number of infinities.
This article is talking about the cardinality of two very specific sets. Before, many believed that t > p, but many believed that this was not provable in ZFC. Both sets contain only sets of integers, so their cardinality is bounded above by the cardinality of the real numbers (the continuum). Some people believed that this result was related to the Continuum hypothesis, and so was only provable one way or the other if you assume the CH or its negative.
As it turns out, both sets have the same cardinality (that of the real numbers) AND it is provable in ZFC.
The title is a little misleading, its really saying that two infinite sets are the same size, but prior to this their cardinality was unknown, and it wasn't even known if the cardinality was provable in ZFC.
There are infinite types of infinity not just two. The result is that they found that two types of infinity that were long assumed to be different turned out to be the same.
This is more interesting than it sounds because these infinities are in between (but possibly equal to one of) the size of the natural numbers and the size of the real numbers. There are hard limits on what can be known about such infinities in the usual mathematical framework (zfc): it's impossible to prove if an infinity which is strictly between the two exists. The hypothesis that no such infinity exists is called "the continuum hypothesis".
Writing this on mobile, but I hope I've made sense...
So does this article prove that two uncountable infinities are equal? Can you tell us precisely what two entities have been proven equal in this article?
I understand aleph-0, aleph-1 and 2^aleph-0 and I also understand that if continuum hypothesis is true, then aleph-1 = 2^aleph-0. Is this what these mathematicians have proven, or have they proven something else?
The paper proves that the smallest cardinality of a set of integers (such that every finite subset has infinite intersections and has no almost-intersections) is equal to the smallest cardinality of a tower.
I use vim and atom, but most of my fellow developers don't want to use atom because it's heavy and slower than st, vim and emacs are out of question for them i think because things like the amount of work needed to prepare all the plugins to get the same functionality as the other tools.
Did you know that there are rich distributions that have done that for you? You can literally install those with a few simple steps and have fully functional super environment.
Key words for Google search: Emacs Prelude, or Spacemacs
It is not that GNU tools were removed. It is more like GNU tools were not added. There is no reason too. Mac provides its own POSIX compliant tools. If you prefer GNU tools, well there is GNU/Linux. If you don't mind more traditional POSIX compliant tools, then macOS is a good option, as are FreeBSD, OpenBSD, etc.
RE: Linux distros, I am familiar with Ubuntu, Debian, and Arch Linux. To some extent RHEL as well.
RE: Unix (commercial brand names), none. It was my fault that I assumed the recruiter meant UNIX-like systems.
I can claim experience with developing userland applications portable across POSIX-compliant systems, as well as some understanding of lower-level operating system concepts. However, to be honest it seems I am unqualified for the role as specified, which makes the exchange all the more embarrassing. Oh well, what's done is done.
The recruiter did mean UNIX-like systems. There exists no useful set of experience which is common among pedantically defined "UNIX" and which excludes UNIX-like systems. The pedantic definition-lawyer type responses you've gotten are absurd.
A role might specifically require extensive AIX experience in particular -- and if it did you would say you needed "AIX experience" -- never some vague umbrella family of "UNIX experience."
It is entirely unbelievable that the requirements for this role would exclude your experience. Doubly so as they're recruiting from interns. Triply so if you know anything about Facebook engineering teams. "Production Engineer?" "Frontend Engineer?" At Facebook? Yeah, those teams run Linux.
ps: The folks telling you not to language lawyer are correct. But your initial assumption that the recruiter had made a dumb mistake was also correct.
The recruiter in this case is a pattern-matching engine with no internal semantics, as a result of Facebook hiring them and training them but not educating them.
Eventually they'll notice that they are only seeing car salesmen with extensive experience in selling brown cars with walnut dashboards, rather than carpenters. [ https://news.ycombinator.com/item?id=15190438 ]
You honestly did nothing wrong. I don't think Facebook is looking for UNIX admins/developers, nor are they going to find any for an internship. CS students with experience in AIX, HP-UX, Solaris, or FreeBSD are exceedingly rare. Solaris and FreeBSD are the only two systems that Facebook would have a chance in hell at finding interns for, and even that's a stretch. AIX and HP-UX are expensive and the licensing is painful.
Getting a little frustrated with resume-screening-by-grep is not at all unreasonable. After the recruit had you play whack-a-mole for three resume revisions, a little bit of snark is to be expected. Don't let people walk over you.
A quick search around the web will tell you that FB probably primarily uses CentOS (with custom built components, I'm sure), so your experience was most likely almost directly what they were looking for.
It would have been pretty simple to tailor your resume in that regard. What I know of PE/SRE type positions at companies like FB and Google is that you are literally working with some Linux flavor nearly all the time unless you are applying for a very niche role.
It is not uncommon to come across open positions that actually do require Unix (not Linux) experience.
For example, American Express has a large deployment of AIX systems and they indeed look for candidates with experience in AIX systems, although I believe someone with experience in another Unix or Unix-like system should also be okay.
But I agree that such openings are usually quite specific about which Unix system they are working with.
This is true, but I would imagine that, if not in the job description, if a really specific kind of Unix experience is required, it would come out in the exchange - and be really OS specific and actually say - True64, AIX, Solaris, etc.
"Unix-like" today means bsd/linux. The recruiter did bad, but recruiters will always do bad.
If its a job you want, don't let a recruiter foible disqualify you, if you can help it.
All the important stuff is in the first 4 pages. I have the PDF version on my desktop. Also printed pages 3 and 4 and stuck it to my cubicle wall.