|
|
Subscribe / Log in / New account

A fork for the time-zone database?

A fork for the time-zone database?

Posted Oct 10, 2021 18:54 UTC (Sun) by marcH (subscriber, #57642)
In reply to: A fork for the time-zone database? by mbunkus
Parent article: A fork for the time-zone database?

> The most universal way is to store just a single name field.

Agreed, but in the Western cultures the family/given split is good enough for the immense majority of people and has endured the test of time and is here to stay.

There have been big changes to family structure and even language recently in the Western world (much more change than in the rest of the world) yet there has been no loud request (yet?) to reconsider this name split. Not even on social media where everyone now loves sharing and spreading the craziest ideas whether they love them or hate them; only the emotions matter.

Even for "foreigners" the main issue is not this split but rather transliteration; a problem much older than computers. I guess that's because families have many different forms but the idea is pretty universal.

> Using "given name" & "family name" is not a solution. ​It just slaps on a different coat of paint on the problem without solving it.

No, it gives the current, unique and good enough solution its accurate... name. Naming things for what they actually are is really not just "slapping a different coat" and in this case it has proved useful for people from various origins who can get very easily and obviously confused by "first" and "last".


to post comments

A fork for the time-zone database?

Posted Oct 10, 2021 19:11 UTC (Sun) by zdzichu (subscriber, #17118) [Link] (20 responses)

I would strongly urge you to read "Falsehoods Programmers Believe About Names" linked in the comments and think about every single item on that list.
There's also a "Falsehoods Programmers believe about Gender" list which contains similar mistakes and eye-opening (or eye-watering for that matter) "Designing forms for gender diversity and inclusion".

A fork for the time-zone database?

Posted Oct 11, 2021 7:42 UTC (Mon) by Wol (subscriber, #4433) [Link] (3 responses)

Exactly.

EVEN IN THE WEST we have two very different naming systems.

We have Christian name(s), patronymic, (surname)
and we have
Christian name(s), (patronymical) surname.

So you can see the drift ... Patronymical cultures don't normally use surname. MANY people don't use First name (the Germans certainly used to use a way of handling this - I've seen Ausweis/Reisepass with the second name underlined, to indicate this is the "name that is used".

I'm guessing that Anglo-Saxon names like Smith and Miller are surnames - typically a *professional* name handed down the generations, while Johnson, Smithson, MacGregor, McIntyre, are patronymics, eg James son of John, Gregor son of Gregor.

When the surname habit froze in I don't know, but I'm guessing it was fairly recent ...

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 11, 2021 9:15 UTC (Mon) by farnz (subscriber, #17727) [Link] (2 responses)

Even in England, we have two systems in parallel; there's the "Christian name, surname" one that most people think of, but there's a significant minority (who have money) that have "inherited name, personal name, surname", where they are only "inherited name, surname" in very formal situations, but they're "personal name" or "personal name, surname" otherwise.

So I have an acquaintance (let's call him "John"), whose name on his passport is something like "James John Smith", but whose bank cards etc are in the name of "Mr John Smith". The only time he uses "James" (which is the name the eldest son was given in each generation, and has the family lands attached to it) is when he's dealing with courts, police, or equivalents.

And this ignores such fun as stage names - e.g. without looking it up, I'd not recognise "Newton Edward" as the first names of someone whose ID normally calls him "Paul Daniels", and yet I know exactly who Paul Daniels is because he was a common fixture on BBC TV 30 to 40 years ago.

A fork for the time-zone database?

Posted Oct 11, 2021 10:15 UTC (Mon) by Wol (subscriber, #4433) [Link]

> And this ignores such fun as stage names - e.g. without looking it up, I'd not recognise "Newton Edward" as the first names of someone whose ID normally calls him "Paul Daniels", and yet I know exactly who Paul Daniels is because he was a common fixture on BBC TV 30 to 40 years ago.

And the reason a lot of people do not use the real name on stage is because acting is a closed shop, you MUST have an Equity card, and no two people can have the same name on active Equity cards.

The reason we know him as David W*A*lliams is because somebody else already had a card in the name of David W*I*lliams.

So yes. It's a very safe bet that most people on stage don't go by their real name because they couldn't get a card with that name on it ...

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 11, 2021 11:36 UTC (Mon) by Wol (subscriber, #4433) [Link]

> The only time he uses "James" (which is the name the eldest son was given in each generation, and has the family lands attached to it) is when he's dealing with courts, police, or equivalents.

:-) To give a real life example, I believe a certain personage uses the name "Charlie Chester" (ONE of his real names!) when he wants to fly under the radar a bit ...

His son was known for a while as Cornet Wales ...

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 11, 2021 8:28 UTC (Mon) by marcH (subscriber, #57642) [Link] (15 responses)

> I would strongly urge you to read "Falsehoods Programmers Believe About Names" linked in the comments and think about every single item on that list.

Pure slideware / bullet points with zero reference, examples, number or any sort of evidence. Zero recommendation or anything anywhere close to a solution.

My favorite falsehood is the last one: "some people don't have a name". Fine, keep not using a computer then and all problems solved!

A fork for the time-zone database?

Posted Oct 11, 2021 11:44 UTC (Mon) by Wol (subscriber, #4433) [Link]

> Pure slideware / bullet points with zero reference, examples, number or any sort of evidence. Zero recommendation or anything anywhere close to a solution.

And why should it set out to provide a solution? It's setting out to provide a map of the bear-traps. It's not setting out to provide a map of your best route around them!

It's down to YOU to find the correct solution for YOUR circumstances. How often do I curse blue bloody murder - AND USUALLY WALK AWY - at websites that - because their designers DIDN'T THINK - won't let me enter correct data. How many websites demand a mobile phone number, when a lot of their customers won't have one? All because they insist on texting you delivery information, for example.

Unfortuntely your typical programmer analyst is doing a job and doesn't bother to think. The analyst is unaware of the real problem-space, and the programmer does what he's told. And the poor sod at the sharp end curses because the problem the computer solves isn't the one he's got!!!

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 11, 2021 13:36 UTC (Mon) by farnz (subscriber, #17727) [Link] (13 responses)

A lot of it comes down to "in this situation, how should the system refer to you?". For example, a chat system might want a display name for other humans to use; a tax system might want an identity the government uses for you. LWN wants nicknames for comments, Facebook wants an "authentic" name.

If you're going to design for a "name", you need to know what you want it for in order to ask the human for the right thing.

A fork for the time-zone database?

Posted Oct 11, 2021 14:37 UTC (Mon) by smurf (subscriber, #17840) [Link] (12 responses)

Exactly. The multitude of address formats causes a whole lot of nonsense data fields to appear in addresses of places which don't need them, or worse slow down delivery because the postal software can't deal. Why the *censored* do some of theses idjits insist on a "state" in my address, and/or auto-infer the name of my neighborhood and insist on adding it to the address in some random location?

I truly don't get why y'all are going to such lengths. Offer a "how do you want to be addressed" field, a "which country / administration do you live in" (that wording should cover Taiwan, Palestine, or Transnistria) menu, and a "what should be on the address label of a package to you" multi-line field. All of which gotta be Unicode capable and not filter out random characters (no, Bobby Tables must not be prevented from causing havoc by filtering user input, thank you very much).

Then you can lean back in comfort and watch everybody else struggle with this nonsense.

A fork for the time-zone database?

Posted Oct 11, 2021 16:45 UTC (Mon) by mpr22 (subscriber, #60784) [Link]

> The multitude of address formats causes a whole lot of nonsense data fields to appear in addresses of places which don't need them

aaaaa

don't talk to me about address formats

I deal with a pair of systems (and their constellation of client websites) that separate "house number or name" from "street", require both fields to be populated, and always present them HNON first

which is a piece of gibbering anglocentric stupidity (and doesn't even work everywhere in england)

but I'm not allowed to set it on fire

A fork for the time-zone database?

Posted Oct 11, 2021 23:09 UTC (Mon) by sammythesnake (guest, #17693) [Link] (10 responses)

Please, NO to free-form addresses.

I deliver for Amazon and every day I have to deliver to addresses that are missing the house number, or the street name, or it's there but spelled incorrectly, or doesn't exist, or have a postcode in a different city from the rest of the address...

All addresses should be either from the database in a canonical form, or behind a scary warning that if you change your address from that canonical form, your package may-or-may-not ever arrive.

Of course, once I get there, I'll still have to handle houses without the number/name anywhere to be seen with numbering orders that must be the product of a stroke* and so on, but let's at least let's solve this one problem that we so easily could, eh?

* Example from today. 10, 9, 3, 4, 7, 8, 5, 6. No house 1 or 2. I swear they get together to drink absinthe and dream up new ways to make no sense...

A fork for the time-zone database?

Posted Oct 12, 2021 0:22 UTC (Tue) by mpr22 (subscriber, #60784) [Link]

Behold the wonders:

https://en.wikipedia.org/wiki/Address

The only internationally portable way to represent addresses, if you don't have the budget to throw piles of money at a company whose primary revenue stream is extracting capitalist rent from control of access to an address database, is as a series of freeform lines written in either Latin letters or the addressee's native script.

It should be noted that even the craziest house numberings usually have some kind of logic, though you may have to analyse its development diachronically rather than synchronically for it to become apparent.

A fork for the time-zone database?

Posted Oct 12, 2021 9:13 UTC (Tue) by farnz (subscriber, #17727) [Link] (3 responses)

Context is everything. If you're getting an address to hand over to a delivery agent, the address should come from a mapping database shared with the agent, so that what you get when you have an Amazon parcel to drop off is in a form usable to you - and that can mean that the address you have for me is a shipping point, not my home (e.g. a PO Box). If you're getting an address to share with the government for tax reasons, it should be in a form that the government is happy with.

And as the delivery company, or the government, you keep a database (either in-house, or bought in from a mapping provider) that contains all the addresses you want in a form that you're happy with. That way, you get a decent location for your parcel deliveries; downside is that your company has to maintain that database so that everyone's address is in it.

Free form addresses are fine if you're gathering something that the user will want to simply share with friends.

A fork for the time-zone database?

Posted Oct 12, 2021 11:51 UTC (Tue) by nye (subscriber, #51576) [Link] (2 responses)

> And as the delivery company, or the government, you keep a database (either in-house, or bought in from a mapping provider) that contains all the addresses you want in a form that you're happy with

That doesn't work in practice though, because the delivery is not done by some nebulous "company", but by a delivery person who has to drive a van around looking for the address on the package. The company hasn't individually looked over each address and checked that it's in the most accessible form; that would be impossible. A person knows their address; "normalising" it to something that conforms to a database created by a machine doesn't magically improve things by virtue of having it in some "consistent" format, because real addresses fundamentally aren't consistent.

My address technically looks like this:
1 Some Street
Town
County
Postcode

Unfortunately "Some Street" isn't accessible by car so isn't on every satnav database, which frequently can't understand that concept. The street that it branches off is on all of those databases though, so after living here a while, we worked out that delivery drivers had a much higher success rate if we render our address as:
1 Some Street
The Other Street
Town
County
Postcode

Systems which try to normalise this end up making the result categorically worse for us and for the people trying to deliver to us.

(Shout out to Kickstarter which requires me to double-confirm my address for every project I back, including having to pick a radio box to say that yes I really do want to use the address you made me re-enter again, and not your oddly mangled version that you describe as "more precise" but has two of the lines merged for no reason. Fuck you.)

A fork for the time-zone database?

Posted Oct 12, 2021 12:22 UTC (Tue) by farnz (subscriber, #17727) [Link] (1 responses)

Which is a failure of the database that's on offer - and really should be in delivery notes, not part of the address. So your address as stored in Amazon's systems should be "1 Some Street, Town, County, Postcode", with a note for the driver saying "Some Street is not accessible by car - go to 2 The Other Street, and then you can walk it".

The delivery notes, of course, are free-form.

Ideally, there would be routes to feed back the issue to the database owner, so that when you pick "1 Some Street" for your Amazon package, the resulting label and bar codes direct their in-van satnav to "2 The Other Street" and tell them to walk the rest of the way.

And Kickstarter have no excuse - when you first add your address to your account, they can ask you if you made a mistake, but after that they should be using the address on file as-is, not insisting on reconfirmation. While they may well be right that you're getting the address "wrong" (if it's a UK address, the correct form loses County nowadays, so it would be "1 Some Street, Town, Postcode"), they should only be asking you to correct it if it's actually causing problems.

A fork for the time-zone database?

Posted Oct 12, 2021 18:52 UTC (Tue) by Wol (subscriber, #4433) [Link]

In the UK the "correct" form often now loses "Town" and replaces it with Post Town 20 miles away ...

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 12, 2021 13:16 UTC (Tue) by jezuch (subscriber, #52988) [Link] (2 responses)

The fix is simple! You need just three boxes: https://what3words.com/

:)

A fork for the time-zone database?

Posted Oct 12, 2021 14:01 UTC (Tue) by excors (subscriber, #95769) [Link]

That's not ideal because What3Words is proprietary and the owners are litigious, plus it appears to be far more prone to confusion between similarly-named nearby addresses than they claim: https://techcrunch.com/2021/04/30/what3words-legal-threat...

A fork for the time-zone database?

Posted Oct 12, 2021 14:19 UTC (Tue) by smurf (subscriber, #17840) [Link]

w3w is categorically bad on literally every metric you'd care to throw at such a system.

Google's plus code isn't that much better.

The only location-encoding system that makes some kind of practical sense IMHO is Uber's H3. It just needs a reasonably pronounceable overlay.

A fork for the time-zone database?

Posted Oct 12, 2021 17:56 UTC (Tue) by Wol (subscriber, #4433) [Link]

> All addresses should be either from the database in a canonical form, or behind a scary warning that if you change your address from that canonical form, your package may-or-may-not ever arrive.

Trouble is, the database isn't in a canonical form ... and all too often they assume that all you NEED is the postcode and the house number. So it's quite normal to have number, street, and the post town 20 miles away ... together with the post code that does uniquely identify the property you want - but it would be nice if the address actually included the town/village where said property was located ...

(Speaking as another guy who has spent the last god knows how many years driving a delivery van ...)

Cheers,
Wol

A fork for the time-zone database?

Posted Oct 23, 2021 14:38 UTC (Sat) by nix (subscriber, #2304) [Link]

Never ever go to Costa Rica. They are slowly gaining a standardized address system, but "the green house half a mile down the road from the place where the gas station used to be" is still commonplace as houses' sole addresses, particularly outside major cities. And you're lucky with that address: it's not "half a mile down the road from the house owned by the Alvarezes"... with an unstated "... and the Alvarezes moved out fifty years ago". I believe it's the single most hilariously impractical addressing system still in common use outside war zones.


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds