Hacker News Reader

Comment by bellyfullofbac | Hacker News Reader

An old story was someone from France wanting to send the Harry Potter book to someone in Russia. They got their address over email, but in the wrong encoding: https://unicodebook.readthedocs.io/definitions.html#mojibake

View on HN

By comparison, recently I bought some small homemade electronics thing from Germany. The sender had written Ålesund on the package. Deutsche Post, or possibly whatever system Posten bought from the lowest bidder, declared that Å isn't a real character, so they dropped it. Unfortunately Lesund is also a place.

After it had been stuck in the sorting facility for three weeks, I called them up and eventually reached a guy who could explain that oh yes, the system does that and the package gets stuck in a loop, but eventually a human will look at it.

It did in fact arrive eventually.

It’d be kind of ridiculous that a national service would not recognize their own characters as a default. It’s more likely that the general-use OCR failed in your instance.

Å isn't a German character, it's Norwegian. GP was ordering a package from Germany to Norway.

But then they mentioned Posten, so whatever the problem was, it wasn’t Deutsche Post’s fault. Also, they should not try to match the address if it is for another country.

If the Norwegian postal service receives an item addressed to LESUND, it's hardly their fault to take it there instead of ÅLESUND (town names are often capitalized).

A post code should have made the error obvious to the postal system.

Right. I might be dense but I understood that the address was written correctly on the parcel.

So whatever the German post thinks the address is, the label should have been scanned within Denmark, and the issue is likely that the Denmark post’s contractor is incompetent, rather than “lol Germans don’t do foreign names with funny letters” (which is not what the OP wrote, but is the vibe in a lot of replies).

And indeed, a post code should render this very difficult in the first place.

The OP wrote that Deutsch Post (probably) or Posten (possibly) messed it up.

The Germans messing it up seems much more likely. There are plenty of Norwegian addresses containing Å/Æ/Ø.

Yeah but that does not make any sense. Deutsche Post would not have changed the label so the address would still be correct.

I have no idea why you think it's inconceivable that Deutsche Post wouldn't corrupt the label, but Posten (Norwegian Post), who actually use the letter Å, would have.

I'm confident OCR fails on a regular basis, which is why they still have humans in place to interpret addresses.

Anecdotally, I had this kind of experience mostly from services in big countries. France, Germany, Italy.

Small countries' services (Slovakia, Bulgaria) need to serve many more customers abroad, so they tend to be more friendly towards foreign characters not found in their national alphabets.

From my experience the French post deals with these just fine. They have to, this issue is quite common. I have regularly received mail with a mangled address because whoever sent it did not bother to check encoding and some Unicode got printed as something else.

Also, if you write the country at the bottom (as you should), they will just send it to wherever that country’s postal services will pick it up without bothering to match whatever’s on top to a local place.

I think the worst mangled address I ever saw was on a package ordered from Italy. The second worst was USA.

Unfortunately, my address has a lot of accentuated characters (Ř, Á, Č, Ž).

I have seen a fair bit of those from the US as well (mine had both é and à).

> Also, if you write the country at the bottom (as you should), they will just send it to wherever that country’s postal services will pick it up without bothering to match whatever’s on top to a local place.

If one is able to design a postal addressing scheme (many just evolve in an ad hoc fashion over time), being able to fail gracefully is a handy feature to have.

In Canada, our postal code system was set up in the 1970s, and is reasonably rational:

* https://en.wikipedia.org/wiki/Postal_codes_in_Canada

If an international shipper can (a) get the package to Canada Post (by writing "Canada" on it), and (b) has the postal code on it, then it can be narrowed down to a fairly small geographic area:

* https://www.google.com/maps/place/A1A+1A1

* https://www.google.com/maps/place/M4A+1A1

That is my experience in the U.K. as well, where it sounds similar to your description. And it works very well, it is a very sensible solution.

I just find it funny that western Europeans can't tell the difference between Cyrillic writing and random gibberish. By which I genuinely don't mean anything pejorative, in either direction.

Do you mean printed, or handwritten? Because Sütterlin looks like gibberish to most Europeans. And it is just the standard Latin alphabet, give or take. It looks like some kind of letters, but they could be from the Voynich manuscript and they would still be as understandable.

In print, it’s partly because several letters are deceptive and don’t sound anything like the closest-looking ones in the Latin alphabet. Some others are just plain weird-looking if you are not familiar with the alphabet already.

> Sütterlin

I figured that the way to keep one's personal communications private is to handwrite them in Sütterlin.

It doesn't help that the cursive Cyrillic T looks like a Latin M.

the most amazing bit is that the postal employees reversed the bad encoding and delivered the package correctly!

Yet it’s sad that they were able to do this, because it suggests that they’ve encountered it enough to develop a procedure for fixing it. (That seems more likely than it being a one-off thing, since I would expect very few people to be capable of figuring out how to convert it, even if they encounter it commonly.)

>That seems more likely than it being a one-off thing, since I would expect very few people to be capable of figuring out how to convert it, even if they encounter it commonly.

I don't know, I'd expect a somewhat tech savvy person in the 90s to figure this out.

>open Word

>type Cyrillic alphabet in upper and lowercase

>try out encodings until the characters look like on the package

>print and use as a translation table

I'm supposing the procedure is the same as it is in software development which is - go ask old Yuri, he knows what to do / about this sort of thing.

And the fact that it was done successfully doesn't mean it was done quickly. Heads off to the Russian dead letter office, someone eventually picks it up for a manual review and thinks "this gibberish is too hard for right before lunch" and puts it back. Next day someone else picks it up and thinks "oh, old Yuri knows what to do with this". Or maybe not, maybe it's done quickly. Who knows.

The slovenian and slovakian embassies meet each month to exchange wrongly addressed mail.

A package sent to me, living in austria, europe, got sent through australia, australia.

Mail errors are common.

It seems strange to me that this is done through diplomatic channels rather than the regular international mail routes between the two countries. Unless you mean the mail for the Slovenian and Slovakian embassies in third countries.

But as another example, Korea Post, the South Korean postal service, recommends including "Seoul" in all addresses from outside the country, even if it's going to a different city, to increase the chances of it going to the right Korea.

Mail for the embassies afaik. and because it is a lot of pretty important mail, they deem it safer to personally exchange it.

Ah that makes sense. Thanks for answering. I was thinking the Slovenian embassy in Slovakia takes all the misaddressed/misrouted mail for the whole country and vice versa.

As long as the postal code is legible, they can forward it to the post office for that area and let someone local figure it out. The rough shape and numbers might be enough to tell without a decoding table.

I guess I am getting old... In the "old" times, computers and software came with manuals.

A common appendix of those manuals were encoding tables, i.e., every column was an encoding and every row a codepoint.

Sometimes you had to introduce relatively common characters by pressing a special key and then the number of the codepoint for the character you wanted to insert.

Back in the day I learned cp437 and its mapping to cp866, and was able to read cyrillic DOS programs without altering the character generator in the graphics card.

This was useful on a computer with a sort of Hercules graphics card which didn’t have the capability in text mode and behaved like IBM MDA.

Needless to say, my knowledge got outdated fairly quickly.

> The address was deciphered by the postal employees and delivered successfully.

Wow, I want to know more about how the postal worker who deciphered it! On a computer, sure just convert it. But on paper, how did they work it out?

I used to have popular books about computers which had character maps of popular encoding in the appendices. Every book about programming, or using DOS/Windows 9x had them.

Each book for system users (not even power users!) had a chapter about different character sets and common problems like this. This was important.

I think the post office worker was enough of a geek or they took the package to the computer guy.

Still more interesting thing to me was that the sender got the address in the wrong encoding and instead of asking via email, just must have thought that Russians have their little ways and dutifully copied all the gibberish (Russians even have a specific term for misencoded characters: krakozyabras).

The Polish term for such characters in Polish is "krzaczki" (little bushes), presumably because of the visual resemblance between those and the misencoded diacritics. It still shows up occasionally.

It's pretty obvious from the letters which are superscripted/skipped

First they knew Russia, Moscow by the postcode, so they superscripted РОССИЯ МОСКВА, and then they knew which garbled letters correspond to these ones

After which they were able to decipher ПР Р.А.СКО.О

after which the address becomes obvious (with the help of index) that it is Проспект Вернадского, 37. I don't think they used any tables other that this implicit table.

Which is student dormitory of Moscow State University btw, I looked it up.

And at no point did the sender stop and think "this doesn't look like Russian". I mean, it's relatively easy for me, who speaks none of the following languages, to tell appart Chinese, Japanese and Korean.

I guess this is consolating for Americans who get blamed for assuming the rest of the world is like their country (postal codes, address formats, date formats, person naming conventions, business customs).

The address there is

Russia, Moscow, (postcode), (maybe a street name), (some number), (untranslated) Svetlane

So it seems to be backwards: The personal name is last, and the country/city is first. Does someone here perhaps know if that is the customary order of address in Russia, or if it was reversed too?

In the whole of USSR (and after its dissolution too), the order of addresses was big-endian: from the larger administrative unit to smaller (which seems quite logical — first you enter the country, then find the district, then the town, then the street, then the house, then the person therein, thus the address replicates the travel path of the package).

Some countries (doubtless those that entered EU, also Ukraine, can’t say for the rest) officially changed the order to be little-endian, but many older folks, who still use snail mail and write letters and postcards, don’t bother.

I think you're mixing up endianness. The way I understand it, your first paragraph actually describes little endian encoding. I understand it as "the little value (i.e., the insignificant value) at the end".

EDIT: In fact maybe I'm wrong, if you consider the lowest byte address the first line of an address label... now I'm confused, haha.

From Wikipedia:

>A big-endian system stores the most significant byte of a word at the smallest memory address and the least significant byte at the largest. A little-endian system, in contrast, stores the least-significant byte at the smallest address.

So if you count the address "fields" from top to bottom in ascending order, putting the country first and street address last would be big-endian. If you put the street address first and the country last, that'd be little-endian.

Yeah it doesn't make that much sense to me either.

I was aware such an ordering was common in China, but thanks for that info about the USSR! It is a more consistent and logical ordering. The other ordering is so confusing that America and Europe have both adopted different mixed endian schemes, which just makes things worse. Consistently increasing-endian schemes are quite rare, but I think you can find one in Australia.

Welcome to the fun that is address field localization. Some countries have house numbers first, and some have it at the end.

742 Evergreen Terrace vs. Platz der Republik 1.

Google's Contacts app just has a text field and lets Google Maps take care of the parsing.

I mean yes, I'm aware different countries do address differently, and that a person should not prematurely predetermine the fields of an address on international software.

I had asked a specific question, which your comment does not address.

I hate and love that. Someone must have seen those characters enough times to have become good at deciphering them.

What a waste of time for the postal system though, I’m surprised they didn’t just return to sender.

Reply on news.ycombinator.com