WTF Bites

Gąska

@Luhmann right on the G key.

Bulb

@Gąska It kind of is, yes. Readability often suffers, too. Nevertheless, off-the-shelf sans-serif look kind of weird in a high-fantasy setting, for example.

That's what font-sets are for. There may not be one one suitable font that covers all the alphabets, but I'm sure they could get separate fonts from the main alphabets (Latin, Cyrillic, Chinese) and complement them with some fallback(s). Maybe Latin could still be custom, with generic period-appropriate Cyrillic, Chinese and such. As long as the texts don't mix the alphabets too much, which they usually won't, it would look OK and work for the non-Latin native writers.

Medinoc

@Bulb This reminds me of the minor annoyance of font fallback on Windows (This may or may not have changed since): if you had a Latin font selected and, say, a Japanese+Latin font, if your text was Latin-Japanese-Latin, the engine would not revert to the original font for the second Latin part.

I wonder if it's still the case, but cba to check.

Applied Mediocrity

@Bulb In vidya you generally want to avoid known unknowns such as font fallback, for performance reasons. Text renderers have come a long way, to a point where you absolutely can throw in all your TTFs and OTFs and expect some DrawString to do all the magic. But glyph caches (even simple sprite maps) are certainly not free.

hungrier

hungrier

@hungrier Update: It eventually failed, one block short. But there's another download!

The same thing happened: It went full speed for 98%, then abruptly stopped and the shown speed descended to

. But then, just as it was about to hit a comically low number of B/s, it shot back up and finished the download!

Which failed to verify.

Jaloopa

@Zerosquare said in WTF Bites:

the best mechanical keyboard ever made

That's on a par with the tastiest shit sandwich ever eaten

MrL

@Jaloopa said in WTF Bites:

@Zerosquare said in WTF Bites:

the best mechanical keyboard ever made

That's on a par with the tastiest shit sandwich ever eaten

JBert

@Jaloopa said in WTF Bites:

@Zerosquare said in WTF Bites:

the best mechanical keyboard ever made

That's on a par with the tastiest shit sandwich ever eaten

Hey, those keyboards do work so take it to the Garage.

Maybe you should have gone for grape jelly sandwiches as an example.

dkf

@JBert said in WTF Bites:

grape jelly sandwiches

I've got homemade raspberry conserve instead.

hungrier

@dkf Not in your keyboard I hope

Zecc

> SELECT AutoIncrementedColumn, ColumnsIActuallyCareAbout FROM TheTable ORDER BY AutoIncrementedColumn

 Yeah, ok, here's your records:

Well, I don't actually need AutoIncrementedColumn, I'm just using it for ordering and pagination.

> SELECT ColumnsIActuallyCareAbout FROM TheTable ORDER BY AutoIncrementedColumn

 Expected column name but instead found AutoIncrementedColumn in SELECT SQL statement

... Seriously?

sebastian.galczynski

New stuff from my very special (as in special ed) clients. Long time ago, they ordered a feature which would pass all their content through Google translate, to generate new language versions of their site. All content, including isolated proper nouns, which the previous crew implemented as requested. What could go wrong?

I quietly (yes, I tried to explain what will go wrong a couple of times, but I gave up) disabled the part that would inevitably cause WTF, but some old data remained. Yesterday they enabled the German locale, and all city and village names are now "translated". We can now go skiing in Begraben, or visit the film festival in Zoo, and many more. The only problem the guy noticed is the missing button which I took away from him, because he wants MORE proper nouns to be translated. So i guess the data is here to stay.

Bulb

@Zecc said in WTF Bites:

 Expected column name but instead found AutoNumberColumn in SELECT SQL statement

Which datascheiße does that?

izzion

@Zecc said in WTF Bites:

> SELECT AutoIncrementedColumn, ColumnsIActuallyCareAbout FROM TheTable ORDER BY AutoIncrementedColumn

 Yeah, ok, here's your records:

Well, I don't actually need AutoIncrementedColumn, I'm just using it for ordering and pagination.

> SELECT ColumnsIActuallyCareAbout FROM TheTable ORDER BY AutoIncrementedColumn

 Expected column name but instead found AutoNumberColumn in SELECT SQL statement

... Seriously?

Not sure if over anonymized, if missing (or extra) comma, or some chintzy sql dialect.

Zecc

@izzion said in WTF Bites:

Not sure if over anonymized, if missing (or extra) comma, or some chintzy sql dialect.

The last one. The order of events wasn't quite like I've described. I spent a good 5 minutes trying to figure out whether I somehow managed to make a typo on a four-letter word. I felt gaslighted (gaslit?) as I rewrote the several variations of the query and sometimes it would accept the column and sometimes it wouldn't; until I managed to pinpoint what the heck was going on.

Edit: re-read the post. I had both AutoIncrementedColumn and AutoNumberColumn. Fixed.

@Bulb said in WTF Bites:

Which datascheiße does that?

I wasn't sure whether I should name and shame DBISAM, but in the end I decided not to.

Bulb

@Zecc said in WTF Bites:

@Bulb said in WTF Bites:

Which datascheiße does that?

DBISAM

WHY THE ? Who had the stellar idea of using such piece of datasheiße?

anonymous234

anonymous234

@Zecc Sounds like you need to SELECT Columns FROM (SELECT AutoIncrementedColumn, Columns FROM ...)

Zecc

@anonymous234 said in WTF Bites:

@Zecc Sounds like you need to SELECT Columns FROM (SELECT AutoIncrementedColumn, Columns FROM ...)

(Ignoring that that's a joke) You think subqueries are possible.

Zecc

@Bulb said in WTF Bites:

@Zecc said in WTF Bites:

DBISAM

WHY THE ? Who had the stellar idea of using such piece of datasheiße?

Decades old third party software.

Zecc

@Zecc said in WTF Bites:

@anonymous234 said in WTF Bites:

@Zecc Sounds like you need to SELECT Columns FROM (SELECT AutoIncrementedColumn, Columns FROM ...)

(Ignoring that that's a joke) You think subqueries are possible.

remembers there's actually a query in another project where something similar is being done

weeps

dkf

@hungrier said in WTF Bites:

@dkf Not in your keyboard I hope

Of course not. That'd be a waste and a sticky mess. Much better to put it in me by means of first putting it on a piece of toast.

Zecc

@dkf No way that post's ending up in QooC. Nope. Can't see it happening.

Zecc

Funny coincidence: these are the two browser tabs I have open

boomzilla

@Zecc said in WTF Bites:

@izzion said in WTF Bites:

Not sure if over anonymized, if missing (or extra) comma, or some chintzy sql dialect.

The last one. The order of events wasn't quite like I've described. I spent a good 5 minutes trying to figure out whether I somehow managed to make a typo on a four-letter word. I felt gaslighted (gaslit?) as I rewrote the several variations of the query and sometimes it would accept the column and sometimes it wouldn't; until I managed to pinpoint what the heck was going on.

Yeah...in general you have to have the order by column being selected if your query is using distinct. These guys cheaped out and apparently do that for any query. Annoying. And of course the error message is misleading because it obviously is a column name, just not one that was selected, which is presumably the only columns they looked at.

sebastian.galczynski

@anonymous234 said in WTF Bites:

Sounds like you need to SELECT Columns FROM (SELECT AutoIncrementedColumn, Columns FROM ...)

Oh god, this reminds me of hacking the PostgreSQL planner with 'WITH...AS' CTEs. They changed that recently, now you have to write WITH MATERIALIZED to do this.

aitap

Spotted in my own code I don't quite remember writing. Originally without a single comment:

frobnicate <- function(X, foo) { # X is a 3-dimensional array
 for (n in foo) {
  # ...snip...
  idx <- bar() # idx is an integer in [1...3]
  keep <- baz() # keep is a logical vector of appropriate length
  # ...snip...
  X <- list(X) # now X is a list containing the array
  X[1 + seq_along(dim(X[[1]]))] <- TRUE # put three TRUEs after the array?
  X[[idx + 1]] <- keep # replace one of the TRUEs with the logical vector??
  # set the length of the list to the length it's supposed to already have???!
  length(X) <- length(dim(X[[1]])) + 1
  X <- do.call(`[`, X) # call the subset operator with arguments from the list,
  # which is equivalent to X[[1]][ X[[2]], X[[3]], X[[4]] ],
  # which expands to X.original[keep, TRUE, TRUE] or X.original[TRUE, keep, TRUE]
  # or X.original[TRUE, TRUE, keep], depending on idx;
  # now X is 3-dimensional array again, and so the loop continues
 }
}

Variable reuse for the win, I guess?

BernieTheBernie

@anonymous234 said in WTF Bites:

"zh_CH" - what's that? Has Switzerland now its own version of Chinese?
Or was my cow-orker Kevin ("en-EN") working on that software, too?

Zerosquare

@sebastian-galczynski said in WTF Bites:

they ordered a feature which would pass all their content through Google translate, to generate new language versions of their site. All content, including isolated proper nouns

Even if they excluded isolated proper nouns, websites that use automatic translation are still a . Either get a proper translation, or don't bother at all. Showing foreign visitors a bunch of machine-generated gibberish is worse than either option.

Zerosquare

@aitap: the Coding Confessions thread is

Kamil Podlesak

@sebastian-galczynski said in WTF Bites:

@anonymous234 said in WTF Bites:

Sounds like you need to SELECT Columns FROM (SELECT AutoIncrementedColumn, Columns FROM ...)

Oh god, this reminds me of hacking the PostgreSQL planner with 'WITH...AS' CTEs. They changed that recently, now you have to write WITH MATERIALIZED to do this.

Actually, you should always try expression indexes with ALTER TABLE/INDEX SET STATISTICS first, then changing coefficients in the configuration (if possible), then fixing content in pg_statictics, and then abusing limitations (or even bugs) in the engine (as a last resort). They will be fixed, eventually.

Of course, it's quite annoying, especially compared to absolute straightforwardness of DB2 SELECTIVITY clause, but sadly Postgresql developers embraced "no developer ever knows what is correct query plan" as a Prime Directive. I have tried to discuss that with some of them and found out that it's impossible, because they never heard about DB2 and the Oracle Way left deep psychological scars.

Kamil Podlesak

@BernieTheBernie said in WTF Bites:

@anonymous234 said in WTF Bites:

"zh_CH" - what's that? Has Switzerland now its own version of Chinese?
Or was my cow-orker Kevin ("en-EN") working on that software, too?

What's wrong with en_EN? It's what Shakespeare used!

Kamil Podlesak

@sebastian-galczynski said in WTF Bites:

New stuff from my very special (as in special ed) clients. Long time ago, they ordered a feature which would pass all their content through Google translate, to generate new language versions of their site. All content, including isolated proper nouns, which the previous crew implemented as requested. What could go wrong?

I quietly (yes, I tried to explain what will go wrong a couple of times, but I gave up) disabled the part that would inevitably cause WTF, but some old data remained. Yesterday they enabled the German locale, and all city and village names are now "translated". We can now go skiing in Begraben, or visit the film festival in Zoo, and many more. The only problem the guy noticed is the missing button which I took away from him, because he wants MORE proper nouns to be translated. So i guess the data is here to stay.

Please, please, submit that as a front page. Or, at the very least, sidebar.

On a more serious note: proper nouns aside, Google Translate is terrible and you should use something different. (for example, I have a good experience with deepl.com, although not for Polish).

hungrier

@Zerosquare said in WTF Bites:

@sebastian-galczynski said in WTF Bites:

they ordered a feature which would pass all their content through Google translate, to generate new language versions of their site. All content, including isolated proper nouns

Even if they excluded isolated proper nouns, websites that use automatic translation are still a . Either get a proper translation, or don't bother at all. Showing foreign visitors a bunch of machine-generated gibberish is worse than either option.

Most likely their browser can do that all by itself, with the added potential benefit of a newer and better translation engine than whatever you would've used whenever you published.

topspin

Captcha from site @Rhywden linked:

Great. Additive or subtractive mixing? Am I disqualified as a human if I ask that question?

So “violet” did not work. It only told me something was wrong, so I assumed it was the captcha but wasn’t sure. Then I tried “purple” and that didn’t work either. Tried to figure out what else could’ve gone wrong or what else to answer, until I realized that they changed the colors in the question the second time around. Got red and blue again the third time and now “purple” worked. Amazing.

topspin

@Zerosquare said in WTF Bites:

@sebastian-galczynski said in WTF Bites:

they ordered a feature which would pass all their content through Google translate, to generate new language versions of their site. All content, including isolated proper nouns

Even if they excluded isolated proper nouns, websites that use automatic translation are still a . Either get a proper translation, or don't bother at all. Showing foreign visitors a bunch of machine-generated gibberish is worse than either option.

I hate nothing more than going to MSDN or cppreference and being presented with machine translated garbage documentation. For fuck’s sake, give me the English site already.
At least cppreference isn’t as stubborn as MSDN. Yesterday I literally had two MSDN tabs open, one of which was nice enough to have a button along the lines of “read in english” while the other one didn’t. And not in two entirely unrelated parts of it, either. They should have had the same options in a sane world. But no, you need to go to the URL bar to edit in en_US there manually.

loopback0

@topspin said in WTF Bites:

Great. Additive or subtractive mixing? Am I disqualified as a human if I ask that question?

Outs you as a pedant, so yes?

@topspin said in WTF Bites:

So “violet” did not work.

I'd have said purple first.

topspin

@loopback0 said in WTF Bites:

@topspin said in WTF Bites:

Great. Additive or subtractive mixing? Am I disqualified as a human if I ask that question?

Outs you as a pedant, so yes?

Obviously yes, but I honestly didn’t know.

cvi

@topspin said in WTF Bites:

For fuck’s sake, give me the English site already.

I've been unsuccessfully exorcising the last remaining traces of sv_SE from a Windows system for the past few years. Case in point, when I updated Vim a few days ago, the update changed the right-click menu entry from "Edit with Vim" to "Redigera med Vim" (or some such shit). FFS. .

(Ok. Minor amendment. Let me keep sv_SE for LC_TIME and whatever the Windows-equivalent of that is.)

HardwareGeek

@loopback0 said in WTF Bites:

@topspin said in WTF Bites:

Great. Additive or subtractive mixing? Am I disqualified as a human if I ask that question?

Outs you as a pedant, so yes?

@topspin said in WTF Bites:

So “violet” did not work.

I'd have said purple first.

The answer is "magenta".

topspin

@HardwareGeek you’re also thinking additive. As the prototypical pedant around here, I’m not surprised.

While I guess you could argue, as I guess @loopback0 would, that “normal” people understand this as subtractive, I predict that’s going to change in the future as more people use image editing and are rather used to RGB-like color switches than CMYK-like.

loopback0

@HardwareGeek said in WTF Bites:

The answer is "magenta".

Seeing as the captcha is unlikely to be going for maximum pedantry pointzzz, it's clearly both technically correct and completely wrong.

Applied Mediocrity

@Zerosquare said in WTF Bites:

Showing foreign visitors a bunch of machine-generated gibberish

Google Play

Zecc

@topspin said in WTF Bites:

Great. Additive or subtractive mixing? Am I disqualified as a human if I ask that question?

Thank you! Peope tend to forget this.

I'm still annoyed that in two separate episodes of the same TV game show they've asked "besides red and blue what is the third primary color" and "what do you get when you mix blue and yellow" with both answers being "green".

HardwareGeek

@topspin said in WTF Bites:

@HardwareGeek you’re also thinking ~~additive~~correctly.

Yes.

As the prototypical pedant around here, I’m not surprised.

Thank you. I really can't claim to be the prototype, though; there was a fine tradition of pedantic dickweedery here before I joined the forum. Nevertheless, I do appreciate the compliment.

HardwareGeek

@loopback0 said in WTF Bites:

both technically correct and completely wrong.

The best kind of correct.

sebastian.galczynski

@Kamil-Podlesak
The original plan was not to show the translation until it's checked by a human. Then that safeguard was dropped, I think because the the price of human checking was too high. Or was it? Somehow when I bill him for hours of adding and removing some useless columns* from the admin dashboard or other inanities, it's not too expensive. Maybe he just couldn't find any German-speaking guy. I have no idea what's going inside his head.

*The more joins (or some complex denormalization/caching) they require, the better. The best ones result in another ticket "Dashboard is slow!!!111", and the oh-so-needed aggregate column is removed again.

As for submitting it to the frontpage - I think I'll wait until I'm not contractually bound with the guy.

kazitor

@topspin said in WTF Bites:

Great. Additive or subtractive mixing?

It’s purple either way.

Though admittedly, calling the subtractive mix anything other than “dark grey” is very charitable indeed…

~~Fun~~ fact time! Mixing colours from red, yellow and blue pigments alone is a very bad idea. Some of the first literature on “primary colours” as something significant recognised that, while they might produce every hue, dedicated pigments would almost always produce a more vivid (“high-chroma”) colour. But at some point painting teachers abandoned the empirical basis that colour theory had garnered, instead bastardising the fundamentals such that these three specific-yet-unspecific colours are MAGICAL, because someone said so.

How crazy that the printing world, which benefits from producing as wide a range of colours with as few primaries as possible, universally uses cyan and magenta rather than blue and red. How crazy that the “split primary palette” that is able to produce a better range of colours than RYB alone just happens to include primaries much closer to cyan and magenta. HMMMMMMM.

As citation for “red and blue pigments produce murky shit,”

@topspin said in WTF Bites:

And if you make the slightest effort to physically describe subtractive mixing in terms of light, it’s trivially obvious why: red pigment strongly reflects long wavelengths only. Blue pigment strongly reflects short wavelengths only. The overlap of strong reflectance is negligible. The result is almost nothing strong reflected back.

cvi

@kazitor said in WTF Bites:

And if you make the slightest effort to physically describe subtractive mixing in terms of light, it’s trivially obvious why: red pigment strongly reflects long wavelengths only. Blue pigment strongly reflects short wavelengths only. The overlap of strong reflectance is negligible. The result is almost nothing strong reflected back.

Though, one would again have to make it slightly more complicated. You say that red pigment reflects in the red end of the spectrum and blue pigment reflects in the blue end. I would then expect a mixture of both pigments to reflect slightly in either. (IMO, the subtractive mixing makes much more intuitive sense when you talk about transmission, where the light passes through both media.)