Zompist Bboard Again

Tropylium wrote: ↑Sun May 17, 2020 5:44 pm

alynnidalar wrote: ↑Fri Apr 24, 2020 8:11 amI've heard that the reason some people don't like tea/coffee is because they're particularly sensitive to tannins. I've wondered if this is why I dislike coffee and black tea so strongly, but on the other hand, I don't mind dark chocolate, which also has lots of tannins.

Trying to steep your tea in slightly less hot water might help. You get proportionally less tannins or similar bitter compounds into the drink this way. It's usually considered a worse problem for green tea — I used to think green tea available in normal grocery stores is of terrible quality before I started taking the steeping instructions saying "90°" or similar seriously — but the same principle should work for black tea too just fine, especially if it's the more roasted flavors you're not a fan of.

Also steeping for less time. I've never been to a restaurant that had prepared tea that wasn't oversteeped. And I've seen people leave the tea bag in while they drink tea. You wouldn't eat well-done steak; you shouldn't steep tea for more than three minutes.

Linguoboy wrote: ↑Tue May 19, 2020 7:04 pm I'm so bloody thick that I only just now figured out that the Tom Robinson song "Can't Keep Away" is about cottaging.

Why did I read that as "conlanging"?

Nortaneous wrote: ↑Tue May 19, 2020 8:33 pmyou shouldn't steep tea for more than three minutes

I do like Very Strong tea also when the occasion is right, and generally prefer it to Very Strong coffee even; but it will need a hefty amount of sugar too to offset the bitterness (won't override it). There's a Middle Eastern variant I like which involves adding the sugar already at the brewing stage, I believe that affects the extraction somewhat. — Punch-in-the-teeth sweetness is a feature as well of course, I'm talking the tea equivalent of coffee that's portrayed in office comics as having a pudding-like consistency / being corrosive to spoons / [garfield_bean_me.jpg].

Tropylium wrote: ↑Wed May 20, 2020 12:54 pmThere's a Middle Eastern variant I like which involves adding the sugar already at the brewing stage,

Middle Eastern variant? That is how my Mom usually makes her tea, and she's not from the Middle East.

Nortaneous wrote: ↑Tue May 19, 2020 8:33 pmAlso steeping for less time. I've never been to a restaurant that had prepared tea that wasn't oversteeped. And I've seen people leave the tea bag in while they drink tea. You wouldn't eat well-done steak; you shouldn't steep tea for more than three minutes.

It's not bad a rule of thumb, but I do go over the three-minute mark all the time. My morning black tea, for instance, I'll steep 3½-4. And the normal rule of thumb for resteeping green tea is to double the time, so if the first steep is two minutes, the next should be four, and so on (though I find it's seldom worth doing a third steeping unless it's something like pu'er where you're tossing out the first steeping anyhow).

But yeah, in general, if you want a stronger result, just use more tea.

Tropylium wrote: ↑Wed May 20, 2020 12:54 pmThere's a Middle Eastern variant I like which involves adding the sugar already at the brewing stage, I believe that affects the extraction somewhat. — Punch-in-the-teeth sweetness is a feature as well of course, I'm talking the tea equivalent of coffee that's portrayed in office comics as having a pudding-like consistency / being corrosive to spoons / [garfield_bean_me.jpg].

And then, of course, there's masala chai, which commonly involves boiling the milk, sugar, and tea leaves together. (I generally use a modified method which involves brewing the tea, then mixing it with milk, sugar, and spices and reheating it.)

Semi-random: There's an interesting scheme in the Netherlands called the 30% ruling where, basically, certain foreign "experts" who come into the Netherlands can have up to 30% of their salary exempt from Dutch income tax for the first five years of their stay. They're the only country I know of in Europe to have such a thing, which is interesting, but cool for people like me.

Normally, you qualify by having resided outside of a boundary of 150 km from the Dutch border for more than 16 months in the 24 months preceding moving to the Netherlands... but Luxembourg's within the boundary! (I think the entirety of Luxembourg actually is within the 150 km limit line... or enough of it, anyways.) There's a second way to qualify, though, and that's by having completed your Ph.D. within the last 12 months and having met the 150 km rule before beginning your Ph.D. And I qualify that way... but I have to show proof documents of my residence before I started my Ph.D. in Luxembourg. Which was almost five years ago. (Fun!) And I'll also need proof of my Ph.D.; I don't yet have the diploma, and won't have it until December, so I'm just hoping the certificate of completion I have will be enough.

Either way, I do have a few months to gather the stuff to apply. But juggling that plus another tax thing I'm trying to get taken care of... yeah. "Fun."

I had the most wonderful dream, one of the longest and most detailed I've ever had. But it looks like the meticulous writing I did was in fact part of another dream.

Even so I was able to work out the tail end of the plot .... of the outer dream anyway ... by starting at the end where I woke up and working backwards. I'll post it soon, probably on the dream thread since I haven't updated my website lately.

Also, I'm moving in July and may be back to mobile hotspot so I won't be posting as much.


coronavirus
cornavirous
carnovirous
carnivorous

PETA was right!!

Pabappa wrote: ↑Thu May 21, 2020 7:58 am
coronavirus
cornavirous
carnovirous
carnivorous

PETA was right!!

ELVIS
LEVIS
LEIVS
LIEVS
LIVES

when i move to my new apartment i might put adhesive rollup paper on the wall and then start drawing on it as if it were a giant whiteboard. its cheaper by far than an ordinary whiteboard and when i did it in this apartment it didnt leave even the slightest mark. plus, i can put up several of them at one time or even cover the whole wall.

what would i draw? probably symbols in my conscript, which exists really only in my head, as I know what the shapes are supposed to look like, but I have never been able to come up with letter assignments that satisfy me. this means that the drawings i make will be abstract too, and "look like writing" because the symbols are all the same size but yet they would be totally unreadable even to me.

i may also do some real writing in Poswa, though, in Roman letters so I know what it means.

Costa Rica just legalised same-sex marriage!
A very big step for the country considering that back in 2016 they nearly elected an Evangelical fascist whose main policy was banning lesbians.

mèþru wrote: ↑Tue May 26, 2020 9:23 am Costa Rica just legalised same-sex marriage!
A very big step for the country considering that back in 2016 they nearly elected an Evangelical fascist whose main policy was banning lesbians.

What was his slogan? "Tough on lesbians, tough on the causes of lesbians"? And why lesbians specifically? Usually lesbians have it a bit easier than gay men when the homophobes are in charge, because they threaten the masculinity of the (usually male) ringleaders less.

So, I have to read some values from a file, filter them, then sort them. Hmm, given that I have to load them one-by-one and that's probably the slowest part, a simple insertion sort should work. Let's make a quick implementation and test it... OK, it works. Now, let's use a real data file... man, this is taking ages!

Well, normally Quicksort is faster. Oh, there's a "qsort" in the stdlib! Well, let's try that. Oh, this has finished in seconds.

Moral of the story: trust standard algorithms.

chris_notts wrote: ↑Tue May 26, 2020 10:58 amUsually lesbians have it a bit easier than gay men when the homophobes are in charge, because they threaten the masculinity of the (usually male) ringleaders less.

And because lesbian sex is "not real sex" (according to them).

Ryusenshi wrote: ↑Tue May 26, 2020 11:20 am So, I have to read some values from a file, filter them, then sort them. Hmm, given that I have to load them one-by-one and that's probably the slowest part, a simple insertion sort should work. Let's make a quick implementation and test it... OK, it works. Now, let's use a real data file... man, this is taking ages!

An ex colleague of mine implemented his own text file based "database" with no adequate backup solution on a server because "SQL databases aren't fast enough". We're not even talking blob storage in a DB here, we're talking literal ASCII text files being reparsed to send back numeric data on every request to the server. The guy was an ass.

In my case this is a purely personal project that probably won't be used ever again, so I'm not sure if it's really important.

Ryusenshi wrote: ↑Tue May 26, 2020 11:45 am In my case this is a purely personal project that probably won't be used ever again, so I'm not sure if it's really important.

I'm not criticising you. In his case this was for a production customer facing system.

Ryusenshi wrote: ↑Tue May 26, 2020 11:20 amSo, I have to read some values from a file, filter them, then sort them. Hmm, given that I have to load them one-by-one and that's probably the slowest part, a simple insertion sort should work. Let's make a quick implementation and test it... OK, it works. Now, let's use a real data file... man, this is taking ages!

Well, normally Quicksort is faster. Oh, there's a "qsort" in the stdlib! Well, let's try that. Oh, this has finished in seconds.

Moral of the story: trust standard algorithms.

Heh. Yeah, it's hard to beat libraries that have had a lot of thought going into them, compared to a little algorithm adapted in five minutes from a reference site.

Hobby experiments aside, I'd say that in the real world it'd be better not to re-sort the whole thing for every added item, because that's too expensive. Maybe just decide on a lapse of time to make a new database index on the more relevant recent part (and store the old parts with an index). Or alternatively decide on a lapse of time to re-sort the thing.

Also, you may be amused to hear the qsort() function in C and C++ standard library implementations is not necessarily the pure Quicksort you find in textbooks, but something fancier that tries to predict which of Quicksort or Mergesort will be faster to execute (depending on the size of the array and maybe the complexity of the comparison function), and that, when choosing Quicksort, samples the array if it's large to get a better chance at choosing good pivots (Quicksort is very vulnerable to bad initial pivots), or when choosing Mergesort, tries to decide whether making array copies or making and following pointers will be faster. The C standard does not require qsort() to use Quicksort.

Also, your insertion sort may have been slower not only because it was a worse algorithm in this case, but because the qsort() implementation you used may have custom code for your processor architecture, like parts written in assembly (or lower) that skip some runtime checks that your C compiler puts in.

Just store them in any order then sort once. Or add them to a (standard implementation of a) heap as you go then pull them off the heap at the end. Heapsort has better worst case behaviour than quicksort.

chris_notts wrote: ↑Tue May 26, 2020 1:31 pmJust store them in any order then sort once. Or add them to a (standard implementation of a) heap as you go then pull them off the heap at the end. Heapsort has better worst case behaviour than quicksort.

Oh, I wrote the above thinking of, say, continuous incoming streams of data (live data) that you have to show sorted to the outside, or some situation like that where you are actually forced to take items one-by-one and regularly re-sort them. If it's a single file, you only need to sort the data once at the end, yeah.

I've seen people store data with CSV files and query it by loading the entire file into a pandas dataframe. Academics, so not people who know SQL -- and even when academics do know SQL they make a complete hash of it -- but, good lord, some things should be part of the standard curriculum. Any reasonably intelligent person with good instruction can learn what a join is in less than a week. Setting up Postgres is a pain in the ass, but we have sqlite for that now.

chris_notts wrote: ↑Tue May 26, 2020 1:31 pm Just store them in any order then sort once.

Yep, that's exactly what I did and it works like a charm.

Nortaneous wrote: ↑Tue May 26, 2020 11:04 pm I've seen people store data with CSV files and query it by loading the entire file into a pandas dataframe.

Guilty as charged.

Nortaneous wrote: ↑Tue May 26, 2020 11:04 pm Any reasonably intelligent person with good instruction can learn what a join is in less than a week. Setting up Postgres is a pain in the ass, but we have sqlite for that now.

I did look into SQL databases for another project, but got kinda lost between all possible versions (SQLite? MySQL? PostgreSQL? MariaDB? LibreOffice Base?). Especially the part when you have to set up a server and a client, even if you're only running on a single computer.