Apple has pushed out an emergency update for all its operating systems and devices, including TVs, watches, tablets, phones and Macs.
The fix patches a widely-publicised vulnerability known officially as CVE-2018-4124, and unofficially as “one character to crash your iPhone”, or “the Telugu bug”.
Telugu is a widely-spoken Indian language with a writing style that is good news for humans, but surprisingly tricky for computers.
This font-rendering complexity seems to have been too much for iOS and macOS, which could be brought to their knees trying to process a Telugu character formed by combining four elements of the Telugu writing system.
In English, individual sounds or syllables are represented by a variable number of letters strung together one after the other, as in the word expeditious.
That’s hard for learners to master, because written words in English don’t divide themselves visually into pronunciation units, and don’t provide any hints as to how the spoken word actually sounds. (You just have to know, somehow, that in this word, –ti– comes out as shh and not as tea.)
But computers can store and reproduce English words really easily, because there are only 26 symbols (if you ignore lower-case letters, the hyphen and that annoying little dingleberry thing called the apostrophe that our written language could so easily do without).
Better yet for computers, English letters always look the same, no matter what other letters they come up against.
We do have some special characters in English typography – so-called ligatures that combine letters that are considered to look ugly or confusing when they turn up next to each other:
These English ligatures aren’t taught in primary school when you learn your alphabet, so you can go through your life as a fluent, native, literate English speaker and not even realise that such niceties exist. It’s never incorrect to write an f followed by an l, and little more than visual politeness to join them together into an fl combo-character.
Additionally, in English, we sometimes pronounce –e– as if it were –a– (typically in geographical names where the old pronunciation has lingered on, as in the River Cherwell in Oxfordshire, which is correctly said aloud as char-well); sometimes as a short –eh-,sometimes as a long –ay-, and sometimes as if it were several Es cartoonishly in a row, –eee-.
But many languages use a written form in which each character is made up of a combination of components that denote how to pronounce it, typically starting with a basic sound and indicating the various modifications that should be applied to it.
So each written character can convey an array of information about what you are looking at: aspirated or not (i.e. with an –h– sound in it), long vowel or short, sounds like –eee– rather than –ay-, and so on.
That’s great when you are reading aloud, but not so great when you’re a computer trying to combine a string of Unicode code points into a visual representation of a character for display.
It’s particularly tricky when you are scrolling through text.
In English, each left-arrow or right-arrow simply moves you one character along in the current line, and one byte along in the current ASCII string, but what if there are four different sub-characters stored in memory to represent the next character that’s displayed?
What if you somehow end up in the middle of a character?
Or what if you split apart a bunch of character components incorrectly, accidentally turning hero into ear hole or this into tat?
For that reason, unusual (or perhaps merely unexpected) combinations of characters sometimes cause much more programmatic trouble that you’d expect, as when six ill-chosen characters brought Apple apps down, back in 2013…
…or the recent CVE-2018-4124, when Macs or iPhones froze up after encountering a message containing four compounded Telugu symbols that rendered as a single character:
(For an intriguing overview of complexities of rendering Telugu script, take a look at Microsoft’s document entitled Developing OpenType fonts for Telugu script.)
The February 2018 “Telugu bug” was particularly annoying because a notification containing the dreaded character could cause the main iOS window to crash and restart, and to crash and restart, and so on.
Unsurprisingly, given the ease of copying and pasting the treacherous “crash character” into a message and sending it to your friends (or, perhaps, your soon-to-be-ex-friends), Apple really needed to get a patch out quickly.
And now it has.
What to do?
For your iPhone, you ‘ll be updating to iOS 11.2.6; for your Mac, you need the macOS High Sierra 10.13.3 Supplemental Update.
To make sure you’re current (or to trigger an update if you aren’t), head to Settings
→ General
→ Software Update
on iOS, and to Apple Menu
→ About This Mac
→ Software Update...
on your Mac.
Oh, and if you will forgive us a moment of sanctimoniousness: if you were one of those people who sent your (ex-)friends a message containing the Telugu bug because you thought it would be hilarious… please don’t do that sort of thing again.
With cyberfriends like you, who needs cyberenemies?
John
*iOS 11.2.6
Paul Ducklin
Fixed, thanks!
MikeP
The apostrophe is essential to distinquish between a plural of a word (such as peoples) and a possessive meaning of a word (such as people’s). It is also helpful in some abreviations.
Paul Ducklin
But the apostrophe is not essential at all – its often helpful, as you say, but we dont vocalise the apostrophe when we talk, yet were easily able to tell the difference between “The Peoples Choice” and “The Peoples of the World”.
Its not even as though we use the confounded thing consistently – theres only one apostrophe in the word “shant”, for example (but by my count there should be three, or even four if you denote the elided space). And in the phrase “The Peoples Choice”, there is an apostrophe, yet it no longer denotes any missing letters at all, because there arent any, so whats that all about?
Indeed, the apostrophe is so often misused that we have a phrase, “The Grocers Apostrophe”, to make fun of people who dont use it correctly – even though we have no trouble at all understanding exactly what they mean when they write “Oranges” instead of “Oranges”, or “its” instead of “its”.
Thered be a few years of adapting to the fact that words like “cant”, “wed” and “were” now have two different pronunciations, to be determined by context. But English is already so littered with absurdities of sound (cough/bough/lough, for example) that Im sure well manage just fine.
When Im King, you can look forward to the apostrophes banishment from my kingdom. (And I shant allow milk or sugar in coffee any more – thats a simple failure of good taste.)
Nobody_Holme
I was wondering your sanity until you suggested touching my coffee.
No. No kingship for you, sir.
Paul Ducklin
Think of it as a Purity Law, but for coffee, not for beer. No one is touching your coffee, only the impertinent adulterants.
🐧 Penguinning 🐧 (@pengtard)
You sound more like a dictator than a king. No sugar or milk in coffee? Banished! We’ll be a republic.
With apostrophes.
Paul Ducklin
Becoming a republic won’t, oops, sorry, wont immunise you from dictatorship. It wont suddenly make it right to put milk in coffee either.
Colin Cohen
‘a fluent, native, literate English speaker and not even realise that such niceties exist. It’s never incorrect to write an f followed by an l, and little more than visual politeness to join them together into an fl combo-character’
I spent my life in printing and publishing ad never have seen it suggested that you might write a ligature. Calligraphers use them for effect all over the place not just for fl, fi, ffl, ffi, ct, st or whatever.
As for Cherwell go back 2-300 years and it is Charwell on maps, otherwise just for yokels and the Oxbridge set!
Paul Ducklin
Ligatures in Roman script, so far as I know, arose from writing, not so much as flourishes but as ways to speed up the process before printing was perfected – thus common Latin endings such as -arum and -orum were replaced with a bar above the root vowel, and the word et (meaning “and”) was squeezed into a shape that became our ampersand (&). However when I said “write”, I sort of meant “when you perform some task that ends up with the written or printed word”.
Quite a few place names seem to have avoided the great vowel shift from E-sounds-like-A to E-sounds-like-E. Cherwell is one (the river flows into the Thames a few kilometres north of Sophos); others that spring to mind are Derbyshire, Berskshire and Clerkenwell.
Colin Cohen
Not at home, so can’t check, but not sure those diacritics are ligatures. Certainly during the incunabula there were loads of alternate sorts for many characters so they could have different widths to more closely resemble script and to aid the justification of lines before the idea of the variable space
Paul Ducklin
I think you are right. Admittedly my intellectual effort ran into (and then ran out at) Wikipedia, where the article entitled “scribal abbreviations” (which it refers to as sigla) covers exactly what I was talking about. I guess in modern parlance you would use the word “shorthand” or “abbreviation”, not “ligature”. Therefore simply ignore enough of what I said earlier to make me non-wrong.
Mike Timms
Sorry Paul,I can’t agree about the apostrophe, it makes life so much easier. I wonder if you speak any other languages where you have to go around the houses to indicate a possessive?
Paul Ducklin
Sure, it helps a bit in written English, in the same way it is useful to have a separate written word for “program” (in the sense of computer software) and “programme” (in the sense of a TV show).
Yet in spoken English they sound identical and we manage just fine. In fact, in American English they are written identically – “program” is used in both cases – and no one trips over that.
In short, the problem with the apostrophe is that it is a huge source of error, yet it is largely useless, and is often used to denote a missing letter that doesn’t exist anyway.
So we might as well leave it behind, just as we stopped writing “ye olde” and standardised on “the old”. (They are pronounced identically, if ever you wondered, because they are the same words written differently.)
Mike Timms.
John’s car or el coche de Juan? I rest my case.
Should you ever make King I shall consider emigration, but have no plans to buy tickets just yet. Keep up the good work.
Paul Ducklin
The weird thing in your example is that if you had written “Johns car” – which is exactly how you say it, without vocalising the apostrophe – there would be no possibility of anyone misunderstanding you. So you are arguing that the English possessive is nice and compact, compared to some other languages, and that’s why it is so cool. Therefore my way, which is even more compact and unemcumbered by pointless dangly bits, is even cooler :-)
English is far from unique in this respect. Many Germanic languages are just like English, using two words only, namely the word for “house” plus the possessive form of the name “John”. Many Romance languages use a preposition to denote the possessive instead – just as English does in the song “The House of the Rising Sun.”
Occam’s razor says, “Occams razor.”