Hacker News Reader

Comment by jrockway | Hacker News Reader

> Apple will offer more price points for apps. Today there are less than 100, but the company is committing to increase that to more than 500 options.

I really hope they implemented this as an enum, so someone has to wade through:

    enum PricePoint {
       UNKNOWN = 0;
       NINTEY_NINE_CENTS = 1;
       ONE_DOLLAR = 2;
       TWENTY_DOLLARS = 3;
       NINETEY_EIGHT_CENTS = 4;
       DEPRECATED_NINETY_SEVEN_CENTS = 5;
       TEN_DOLLARS_PER_MONTH = 6;
       PORTUGAL_NATIONAL_BALLOON_DAY_2019_SPECIAL_NINTEY_NINE_CENTS = 7;
       // 500 lines follow.
    }

View on HN

Lol, few things trigger my OCD more than misspelled words in identifiers. I legit missed important use case while amending legacy system just because someone decided to have “priveleges” in the variable name. I hope all those ways to spell number 90 are part of the joke.

Oh yeah it's especially wonderful if it's too late to fix the spelling mistake lol.

I feel like being proactive about these things is one of the most important parts of my job.

I'll just leave this here for you:

https://datatracker.ietf.org/doc/html/rfc7231#section-5.5.2

That's a good one. Since legacy code is less likely to change than the English language, I just call it a "referer" in writing unrelated to HTTP. That way, I'm never confused. (Everyone else? Good luck.)

I once saw this implemented as a regex. Unfortunately the regex was refer*er.

“Tests pass, ship it”

The tests clearly did not include spinning up an HTTP server and actually using the correctly-spelled header (as per correct English), then.

I don't know enough about overflows, but that could theoretically be exploited, right? referrrrrrrrrrrrrrrrrrrrrrrrrrrer

inb4 refeer (should've used referr?er)

I also feel like this is a buffer overflow waiting to happen. If somehow someone makes the header length way longer than expected (eg. referrrrrrrrrrrrrrer).

? is 0 or 1 instances of the preceding character, so referrrrrrrrrrrrrrer wouldn’t match /referr?er/, only referer or referrer.

Proof: https://regex101.com/r/s56ebD/1

I'm replying to the grandparent comment. I said "Also" meaning I was adding onto your comment.

I used to do technical writing and for some software that I wrote the manual and help system for, one of the menu items was misspelled. I spelled the word correctly in the text describing its functionality, but whenever I referred (so to speak) to it directly as the menu item to be selected, I kept the misspelling (this was the early 90s, the days of people saying non-ironically that they couldn't find the "any" key on their computer).

Pretty sure people still can't find the any key, we just stopped asking them to!

Yep, I get some prompts now that say "Press the $X key" (often space), but will actually proceed if you hit any key. It always give me a touch of chuckles when I think about why they named a specific key then.

Lets not forget the misspelling of guillemet as guillemot that is now firmly embedded into PostScript and Adobe software with no chance of ever being fixed.

Unicode sidestepped the issue by naming the characters RIGHT/LEFT-POINTING DOUBLE ANGLE QUOTATION MARK.

Unicode also documents its mistakes in technote 27 “Known Anomalies in Unicode Character Names” (https://unicode.org/notes/tn27/), with remarks such as

“U+034F COMBINING GRAPHEME JOINER

The name does not describe the function of this character. Despite its name, it does not join graphemes”

“The use of the spelling lamda derives from ISO 10646. This does not mean that it is more correct than lambda, merely that the spelling without the 'b' is the one used in the formal character names”

“A spelling error: "brakcet" should be "bracket". A formal alias correcting this error has been defined”

Why can't unicode change the names of the glyphs?

From that page:

“Because Unicode Standard is a character encoding standard and not the Universal Encyclopedia of Writing Systems and Character Identity, the stability and uniqueness of published character names is far more important than the correctness of the name. The published character names are normative for the purposes of the Unicode standard and the large number of other IT standards that reference it. These standards require stable identifiers and character names must therefore be immutable — any change of character names is almost as disruptive of the standards as changing code points for characters would be. Accordingly, the Unicode Consortium has adopted the Name Stability Policy, preventing changes in character names. As a result, errors in character names cannot be corrected. Instead, important character name anomalies anomalies (SIC) are documented with annotations in the Unicode Character Code Charts.”

Also because somebody, somewhere, might have used those names to look up Unicode code points.

Part of each Unicode release is a file named “UnicodeData.txt” (https://unicode.org/Public/13.0.0/ucd/UnicodeData.txt), so those strings are public information.

Just a note in passing: Unicode doesn't name glyphs; it names characters.

I’ve heard Greeks insist that lamda is indeed correct (as a non-misleading approximation of something like [lamða], not necessarily as an English spelling), so that part might have been a deliberate result of the ISO process.

(There’s also the name caron for the haček, apparently not used by anyone before the 80s; per the official FAQ:

Q: Why is the haček accent called “caron” in Unicode?

A: Nobody knows.)

Not for lack of trying: https://www.babelstone.co.uk/Blog/2009/08/antedating-caron.h...

The key is to write out the enums to long-term storage as the names instead of the numbers, that way nobody can ever fix your spelling mistakes.

There are lots of ways for spelling errors to get reified. They can also end up in a data interchange XML schema (I bumped into that one at a previous job where it seemed like an obvious and trivial fix to correct the spelling of a class name and discovered that it had ended up in a company-wide data interchange schema which could not be practically fixed).

The most important example of that class of error is REFERER. It should have been spelled REFERRER. But nobody caught it, and you really can't change something that is part of the HTTP standard...

Early header compression.

Took a leaf out of the UNIX creat/umount handbook, clearly.

umount() is actually legit - early C limited extern identifiers to 6 chars (or rather it would allow more than that, but the rest would be silently ignored for linking purposes), hence why we also have strlen() etc.

creat() is weird because create() would fit that constraint just fine, so there's no obvious reason as to why it had to be shortened. The 1978 K&R C book actually notes it with [sic].

When you're dumping loads of short, generic, common words into the global namespace (well ... the only namespace in C), it's nice to have them misspelled so user code is still free to use the real word.

> well ... the only namespace in C

Technically there are several namespaces since you can, for example, name a struct the name as its type.

After many years in Javaworld, seeing this sort of thing in C++ really freaked me out.

At this point I think it may be question whether the problem lies with English itself. Such an error prone language.

You don't know the half of it. Many contemporary and common words are the results of misspellings.

“a nickname” was originally “an ekename” (eke being an old English word meaning also)

conversely, “an apron” was originally “a napron.”

English is a weird agglomeration of different languages, with things like the application of French-style plurals (-s) to Anglo-Saxon words (but not always, cf. ox -> oxen) and a generally Germanic syntax with a large fraction of Romance root words.

As part of my Dewey Decimal Project¹ https://www.dahosek.com/category/dewey-decimal-project/ I ended up reading some interesting books on English linguistics (scroll down to the 500s for those).

———

1. I’m not quite done. I have two read books to blog about before I read the last book in the series which will be the book my local library has at its highest Dewey Decimal call number. Alas, it's not a 999 Extraterrestial Worlds book, but I think it's either Antarctica or Australia (I looked when I first started on this seven years ago, but I've since forgotten).

Also, "a narang" -> "an orange".

It's not really misspellings, it's reanalysis, and it's not at all particular to English.

Tangent: this reminds me of the _aptly_ named "mod_speling" module.

> The most important example of that class of error is REFERER. It should have been spelled REFERRER.

I thought it was supposed to be REEFERER.

Or get your typo into a spec like "referer"

I was just trying to debug something today and noticed how the referer HTTP header is spelled wrong and that is the spec while there are several other headers that spell it referrer. Crazy!

And everyone saying "backslash" is on the inside of an awesome joke too.

Have fun naming the variable you use for referer.

do you actually have OCD or are you just using this as a figure of speech?

As a dev with dyslexia, I apologize. I've for sure committed some sins there in places that are nasty to fix.

Well, pasting the enum example into IntelliJ highlighted the spelling mistakes and suggested replacing NINTEY and NINETEY with NINETY. I think this is how my dyslexic colleague works too - copy pasting instead of writing and looking for various wiggly lines in code editor.

Yeah, spell check in editor is killer, and I work in strongly typed languages so it's a compiler error if I misspell a variable name is another big one.

It doesn't catch everything however.

In a triple A game I worked on, the PlayStation button enums were CROSS, SQUARE, TRIANGLE, and...... CRIRCLE.

Would drive me absolutely livid. But of course no one was allowed to fix is since it was already in a few different games, countless script files, network-controlled configs etc.....

I feel your pain. A company I work for has a weird name that doesn't follow the rules of English spelling correctly (I can't say more without giving away the company name). This was an intentional decision and poses enough of a problem for guests visiting the company. But the absurd thing is this company decides to double down on it's troubles by naming all of it's projects using the same broken rule as the company title. Often these projects will be common English words re-spelt in the style of the company name. So now I'm often finding myself misspelling common words because of this stupid company name. And all of this is intentional too! Drives me nuts.

Googol?

At this point most students learning about that number assume it's misspelled when first introduced and incorrect the teacher.

>>incorrect the teacher.

Did you just use incorrect as a verb?

A favorite pastime of HN is users incorrecting each other on topics they know nothing about. ;)

Reminds me of this comic.

https://www.gocomics.com/getfuzzy/2003/04/27

Verbing weirds language.

Beaten by ten years: https://www.gocomics.com/calvinandhobbes/1993/01/25 :)

That's so discorrect.

unpossible!

postposterous!

I think we're way beyond postposterous. Think postpostposterous.

Reminds me of the occasional encounter with "deinstall" versus "uninstall."

That distinction is a little unrelevant, don't you think? Irregardlessly, you make a good point.

I see what you did here...

> it's troubles

> it's projects

> So now I'm often finding myself misspelling common words

> And all of this is intentional too!

:-D

You mean like Optimizely? I thought most startups these days had absurd adjectiverbnoun names.

CROSS = 1

CRIRCLE = 2

CIRCLE = CRIRCLE

A very long time ago we outsourced our Wordpress development and customization to a developer we found on a website who was from Moldova. He did pretty decent work and we got a great deal, but the HTML/CSS was a bane to work with. Almost half the DIV id's and such were spelled slightly wrong. Didn't affect the operation of the site but was very annoying to work with.

I remember Allegro used to have double functions for Palette and Pallet.

  You may notice that a lot of the code in Allegro spells 'palette' as 'pallete'. This is because the headers from my old Mark Williams compiler on the Atari spelt it with two l's, so that is what I'm used to.

That's what code reviews are for :)

This reminds me when I was learning to code I was working through the Project Euler problems in python, one of them has you write code to do something with the English words for numbers.

I wrote my code and was fairly confident it was correct, submit the answer and it is rejected, spend the next few hours trying to find the logic error. Of course it's because English is ridiculous and it's "forty" instead of "fourty".

What about twoty, threety, fivety and eightty?

I think it's the amount of prefix that gets me with forty. Plus fifth / fourth versus fifty / forty

In an old code base I worked in, someone misspelled “params” as “parms.” Always made me want a sandwich.

Sounds like a lisper.

https://duckduckgo.com/?q=site%3Apaulgraham.com+parms

I worked with a (home-rolled at the company, but long before my time with no compilation documentation) comms library that had functions named "receive" and functions named "recieve".

Even better, the library was called into as a COM object by a Visual Basic Script script, and the misspelled corner case only got hit rarely. Figuring out that the problem was the correctly spelled function call was a joy.

I rarely work with Excel. A couple of days ago I had the joy to need to apply a function with multiple parameters. The formula parser complained with an error message which meant nothing to me. I finally figured out that I needed to use semicolons as separator for the function arguments (I am from Germany). Who in his right mind does THIS? I tried to change the separator in Windows setting, to no avail.

Magento 1.9X used to have a typo in the database that lasted for years because so much stuff already relied on it and it wasn't user facing. I do not miss those days.

You're going to love this: U+FE18 ︘ PRESENTATION FORM FOR VERTICAL RIGHT WHITE LENTICULAR BRAKCET: bracket is spelled incorrectly.

https://en.wikipedia.org/wiki/Unicode#Anomalies

I once fat-fingered a `rails new APPNAME` command and didn't catch it for several months. Not impossible to change, but also not really critical either.

Even worse than misspelled is when it’s inverted polarity! Like uninitialized=true means that it IS initialized. I’ve seen something similar to that in multiple jobs.

There is some evidence of this - if you can log in to app store connect (need a developer account) there is this page:

https://appstoreconnect.apple.com/apps/pricingmatrix/1539646...

It shows the tiers from 0 (free) to 87 (USD 999).

Then come the party tiers: 510, 530, 550, 560, 570, 580, 590 which are just repeats of the other tiers, probably for some good historical reason which I'm not aware of.

> which are just repeats of the other tiers

Those are the price adjusted tiers. They look like repeats in USD, but if you look at other currencies you’ll see they’re different. They’re designed to allow you to offer price points that are scaled according to the market (i.e, 99c in the US, but considerably cheaper than that in India or China).

Makes sense, thanks.

Just note, that link only works for you; if you have a developer account, you'll have to substitute the ten-digit number for one from your apps.

Huh, so $999 is the max -- and here I was all this time thinking the I Am Rich [0] dev had generously chosen an affordable price point.

[0] https://en.wikipedia.org/wiki/I_Am_Rich

You kid but they are a list of “tiers” that get localized.

I kind of assumed. You can't just take the price in dollars and convert it with the exchange rate, because customers don't want to buy a "89.398723 yen" app. It has to be $0.99, £0.89, ¥99, etc.

It is interesting that things are not priced in "buying power", but in numbers that aim to most deceive. ("It's not even a dollar!")

Steam is very often priced in buying power. Depending on the country you live in, prices can be half as high as in other countries.

Why don’t more digital platforms take this approach? Are they worried people from higher relative income countries will log in with a VPN and buy it cheaper?

There’s also the often forgotten fact that in many countries the listed prices include VAT while in others they don’t.

Today I found out we have a national balloon day!

THREE_FIDDY

getOut.gif

* TREE_FIDDY

Supports lookup in O(fiddy).

Complete with spelling errors, semicolons in place of commas, shouting snake case, and unnecessary explicit numbering. Definitely enterprise-grade.

Not complete until it has a poorly documented API with some name like PriceKit, maybe hidden behind an entitlement as well.

The explicit numbering is so that you can reorder them to be more "readable" later. I didn't have the heart to do that here, but in the real world, anything goes!

When somebody comes along and inserts a new case next to an existing case because they're closely related, you'll be glad the numbers are explicit.

Negative - if these are going to be treated like keys where the integer value needs to retain the same meaning between builds, then an enum is the wrong abstraction to use.

The real enterprise case is when additions are made for things that aren’t even pricing

BLUE_BANANA_CONCESSION = 9001

There's also less enterprisy and more "1990s engineering" variant of this:

  enum PricePoint {
    //all the 500+ price points defined above...

    REGULAR_PRICE = 1023;
    PROMO1_PRICE = 1047552;
    PROMO2_PRICE = 1072693248;
    AUTOHINT_ROUND_99 = 1073741824;
    AUTOHINT_BUYING_POWER = 2147483648;
  }

Little bit twiddling, and you can encode 3 prices + some flags in an uint32.

Reply on news.ycombinator.com