randomswede

3 年前議論

Some countries selectively put some people in the "you have a very long notice period" and "you will not be working for us, but still be paid, during your notice period" (so-called gardening leave).

Not unusual in the finance industry, somewhat unusual (but I have heard of it) in "more pure tech". Probably also more common the further up you get in the corporate hierarchy.

randomswede

3 年前議論

In Sweden, at many places that do a materials science engineering degree, it is historically under "Mining engineering" rather than "machine engineering".

randomswede

3 年前議論

The speed of light in vacuum is a hard upper limit. Most signal paths will be dominated by fibre optics (about 70% of C) and switching (adding more delay).

But, yes TrueTime will not magically allow data to propagate at faster-than-light speeds.

randomswede

3 年前議論

My take-away from interviewing is that as a candidate, getting a "you did not get the job" doesn't change anything. Since I've had the privilege of mostly being employed while interviewing for new jobs, this is a "meh, status quo" thing. Nothing changes, I have no decisions to make. If I get an offer, I have a decision to make.

Does it sting when I get a "No"? Yes, a little, but I did my best and (presumably) someone else did better. So, I take solace in that I did not have to make a (relatively large) decision.

randomswede

3 年前議論

I am cusrious what you mean by "LeetCode-style". My understanding is "present a problem, only take the resulting code, judge that".

I did a fair number of coding interviews (for SRE positions) at Google (I don't actually know how many in total, but 25-30 is probably a safe guess) and, yes, it started with a small problem that should be solvable in well under half the interview time. I only used problems that passed my "is this fair to the candidate" screening (look at problem, try to write a working solution, if that takes more than 7 minutes, the problem is not fair).

The value in the interview comes from (a) listening to the candidate narrating their thought process during the coding, (ii) discussing the solution with the candidate, sometimes in terms of complexity, sometimes in terms of correctness, depending on what is on the board, (3) refining the code written (add more functionality, change functionality).

For a language I knew, I tended to overlook small syntactic mistakes (a whiteboard does not have any syntax highlighting) and if there was any questions about library functions, I would give an answer (and note down what I said, so that would be the standard used when judging) and the bulk of the score came from the discussion about the code, not the code itself.

If that matches what you mean by "LeetCode-style interview", that's certainly been in place since at least 2011. If it does not, things may have changed, but at least the aim wit ha coding interview back when I was still at Google was less "get some code judge, on the code" and more "get some code, judge the discussion about the code". It is also entirely possible that interviewing for SWE positions was different frmo hiring for SRE positions.

randomswede

3 年前議論

Even so, it's probably not going to be super-effective, unless each workplace is willing to have enough useful small things that do not require knowing substantial chunks of the existing code base.

It is probably realistic to expect someone to write something useful (although possibly small) in three days. It is less realistic to expect someone to write a useful component integrated in a large system that they have to learn, in three days.

randomswede

3 年前議論

In an ideal situation, you'd get all "planned maintenance" emails for things you care about and no emails for the rest of them.

That (probably) means that the system for dealing with planning maintenances (well, usually, "approving them") needs to have a sufficiently good understanding of what humans care about what changes.

At a previous job, the planned change tracking system was REALLY good at tracking what specific compute facility was going to be impacted by any specific change taking place in that facility. And had a really good way of allowing you to filter for "only places I have stuff running" (and I think, even some breakdown of general change types as well).

It was, however, not easy to get notification of "there will be maintenance on submarine cable C, taking it off-line for 4 hours" or "there will be maintenance at cable station CS, taking cables C1, C2, and C3 down for 3h". And as one of the things "we" (the team i worked in then) was doing was world-wide low latency replication of data, we did actually care that cable C was going to be down. But, the only way we could find out was "read all upcoming changes" and stick them in the team calendar.

Was it good? Eh, it worked. Was it the best process I've seen? Probably ,yes.

randomswede

4 年前議論

If you aim to be "human-friendly" (and that is, as I understand, the raison d'etre for YAML), there is a subtle semantic difference between "true" and "on" (and "false" and "off") and as a human it may be nice to express that semantic difference.

As for that semantic difference, if we expect the light source to have one of exactly two states (that is, "not a dimmable light"), we probably want to express that as "lightsource: on" rather than "lightsource: true".

And that is where the friction between "humanfriendly" and "computer-friendly" starts being problematic. Computer-to-computer protocols should be painfully strict and non-ambiguous, human-to-computer should be as adapted as they can to humans, erring on "expressive" rather than "strict".

I am also not sure if I am happy or sad that the set of configuration languages in the original article didn't include Flabbergast[1], which was heavily inspired by what may be simultaneously the best and worst configuration language I have seen, BCL (a language that I once was very relieved to never have to see again, and nine months later missed so INCREDIBLY much, because all the other ones are sadly more horrible).

[1] https://www.flabbergast.org/

randomswede

4 年前議論

How many seconds in an hour? Most of the time, 3600, occasionally 3601, and very rarely 3599. Hours in a day? Mostly 24, but 23 once a year ad 25 once a year.

These all seem like good reasons to make then functions (taking a timestamp as a n argument) rather than mostly-correct constants.

I swear, the more I learn about calendars and timekeeping, the more I realise I never ever want to deal with it.

randomswede

4 年前議論

Same thing for the ZX-80 and ZX-81. If you were careful with your text prompts, you could save several bytes by injecting a keyword instead of typing it out, but anything that was a "start of line only" could be tricky injecting into a string.

randomswede

4 年前議論

Werrrrlllllll... It all depends on what yo umean by "type". In physics problems, I find taht quantities may be measured in "floats", but have the types "mass", "amperage", "distance per second per second" and the like.

In Go, I would probably model this as:

    type mass float64
    type speed float64
    type acceleration float64

That way, they'd all have float64 as the "storage type", but it would be harder to accidentally pass an acceleration when I wanted a mass.

randomswede

4 年前議論

There is a whole lot of pesky athmospheric interference limiting how much detail you can actually resolve.

randomswede

4 年前議論

These days, my main use of dd is to get a specific amount of data from a file, where both "bs" and "count" are useful (no, "bs" does not only set the buffer, it also sets the chunk size for reads and writes, this is SOMETIMEs useful and/or necessary with tapes).

So, this is an approximation of a command pipeline I run several times per year, when I happen to need a secret of an approximate length:

    dd if=/dev/urandom bs=6 count=1 | base64

Tune the "bs=6", depending on how long you want your (guaranteed typeable) secret to be. Every 3 bytes in the input will give you 4 characters in the output and keeping the input block size a multiple of 3 avoids having the output ending in "=".

It MAY be possible to replace this use of dd with other shell commands. But, since I needed to learn enough of dd to cope with tapes, I use that.

randomswede

4 年前議論

ON the international currency market, one traditional role of the US dollar is exactly "store value". You trade your currencies on the FX market during your day, and comes close to "close of your trading", you exchange all your positions for US dollars, to keep in store for your quiet period. Comes the following day, you trade out your dollars for other currencies. Then the cycle continues.

Does this mean that the US dollar is a terrible currency? No. If another currency meets the same stability criterion, it can (and probably is) used as a short-to-medium term value store.

I don't really see any crypto-currency work well, they're thinly traded, so there's a lot of inherent volatility in the pricing. It is also getting increasingly harder to exchange crypto-currencies for more traditional currencies.

randomswede

4 年前議論

If you are selling A as A-prime, you can be a snake-oil salesman even if you are supremely skilled at A.

In Lars' case, he is from what I can see a supremely skilled instinctive archer. There are some historical documentation, from some regions ,documenting feats that Lars is capable of doing.

But, "feating" and "combat skills" are very different things (well, they are for sword disciplines, I will blithely assert the same is true in archery). Yes, both require skill. Yes, training one discipline can improve the other. but they're not the same. Just like how writing Haskell (or Lisp) can make you a better C programmer.

randomswede

4 年前議論

That'd depend on if they're the same entity as "The Billing Project LLC", that seems to own the "KILL BILL" mark for downloadable software. I hope they are the same.

randomswede

4 年前議論

Oh, boy, on the "never log expected failures as errors" front, I once worked with a database system that used opportunistic transactions. Basically, each modification to a row carried effectively the original value of that trow with the update and if it failed, the API call triggered an error saying that the transaction failed. So if you did a "SET column=(column+1) WHERE rowid=unique", the client could basically do an automatic retry.

But, it also logged each and every occurrence of this at "Error" severity, instead of at "Info" severity (it is, after all, expected to happen once in a while).

And of course, once our code switched over to using this, the first few times every team member had to deal with a production issue, the immediate reaction was "oh, no, the data store is unhealthy! look at this mass of error logs, I can see one every few minutes!". Thankfully, after the first team member (me, as it happens) spent half an hour reading the relevant parts of the design and implementation docs, we could frequently short-cut a lengthy investigation by "oh, you think $DB is bad because you are seeing transaction failures? no, that's expected, see $URL".

randomswede

4 年前議論

Then there's the delightful (no, I actually mean the opposite) errors that g++ emitted (back when I last wrote C++ and compiled using g++), where I basically could go "OK, there is an error that was detected at line L, in file F; and I think it may be a type error", so a recompile with clang, so I can actually understand what the error was, so I could fix it.

randomswede

4 年前議論

Likewise Digital, all VMS error codes are "SUB-S-NAME" (so a permissions error when dealing with a file woule be something like "%RMS-E-NOPERM" (I think that's "rotating mass storage" rather than "Richard M. Stallman"). I think that even harks back to various OSes on the PDP-11, but cannot say for sure.

randomswede

4 年前議論

My (very rough) estimate is that somewhere between 1% and 5% of those who enthuse about crypto-currencies are actually interested in the underlying tech. The rest being in it for one of "it's a safe way to do nefarious economy" and/or "it's easy money".

Me? I think the tech is intellectually interesting, but doesn't really scale to the point where it can provide a "world economy". If nothing else, with Proof-of-Work, you need 51% of ALL available computing resources dedicated to maintaining the chain, which is an incredible waste (if you don't have a majority of all available computing power, the chain is vulnerable). With Proof-of-Stake, you need a majority of the economic system dedicated to just maintaining the economic system.

So, intellectually interesting, but practically useless.

randomswede

投稿

Russ Cox on Memory Models

コメント