The region codesXA andXB in our region bundle don't look like countries. We should check them for mistakes.

But otherwise, I agree about this renaming, as this class deals with a subset of ICU regions, not with all regions, and so is a different concept.

You're right, i suggest to consistently ignore all "non" valid codes, as done in#28833 for languages

• Countries:http://unicode.org/reports/tr35/#unicode_region_subtag_validity
  • excludeXA, XB
  • ICU already excludes deprecated codes:https://github.com/unicode-org/icu/blob/release-64-2/icu4c/source/data/region/en.txt#L302
• Scripts:http://unicode.org/reports/tr35/#unicode_script_subtag_validity
  • excludeZzzz

See alsohttp://unicode.org/reports/tr35/#Unknown_or_Invalid_Identifiers (we almost conform)

Do we agree excluding data is a new feature still?

Thank you@ro0NL.

I think we should revert this change. Regions are not the same as countries. There are lots ofterritorial disputes in the world which (in some cases) makes it impossible to define what a country is.

Even ICU itself declares that:

[Regions are ...] typically codes for countries, but also include areas that are not separate countries, such as the code "AQ" for Antarctica or the code "HK" for Hong Kong (SAR China). Overseas dependencies of countries may or may not have separate codes. The codes are typically 2-letter codes, but BCP47 allows for the use of 3-digit codes in the future.

Also, all the smart developers out there in all programming languages use "Region". You can also see that in all operating systems. Even when displaying it for end-users, they choose "region" over "country":

@javiereguiluz we use theRegions api forCountryValidator andCountrType, that's inconsistent naming at least. This fixesthat inconsistency.

AFAIK the current data are true "countries", as we filter upstream ICU data. As such; did you find a case where a "2 letter region code" is NOT an expected country code?

if so it should be patched IMHO (exclude those), see e.g.#31365

in general i think it's valid to provide a subset of regions; thus countries only.

@ro0NL yes, I've found lots of errors :(

• "AQ": "Antarctica": Antarctica is a continent, not a country:https://en.wikipedia.org/wiki/Antarctica
• "PS": "Palestinian Territories": Palestine is not a country yet:https://en.wikipedia.org/wiki/Palestinian_territories
• "TF": "French Southern Territories": a territory, not a country:https://en.wikipedia.org/wiki/French_Southern_and_Antarctic_Lands
• "TW": "Taiwan": China says that Taiwan is China, not a country:https://en.wikipedia.org/wiki/Taiwan
• "PR": "Puerto Rico": not really a country: it's "complicated":https://en.wikipedia.org/wiki/Puerto_Rico

There are many other mistakes. The only reasonable way to fix all this is ... reverting this PR and using "Region" instead of "Country". That's what everybody else does. Please, consider a revert 🙏 Thanks!

That's what everybody else does.

from a form perspective, not by Symfony:CountryType

Is it aRegionType actually? im really tempted to double check the list and exclude invalid countries such as Antarctica, etc. :/

but then again, renaming everything to Region is also an option. But it's a bigger step for the codebase in terms of deprecation etc.

And if we stick with "regions", doeshttps://github.com/symfony/symfony/blob/master/src/Symfony/Component/Intl/Data/Generator/RegionDataGenerator.php#L43 still apply? Technically we dont care anymore 🤔

i think we should filter byhttps://en.wikipedia.org/wiki/List_of_ISO_3166_country_codes

which statesAQ (Antarctica) is an official (perhaps special) ISO country code

The ISO 3166 country name Antarctica comprises the continent of Antarctica and all land and ice shelves south of the 60th parallel south.

TW (Taiwan) is disputed following ISO-3316, not sure that makes it an invalid country. I would keep it for now.

So following ISO-3316 seems like a good compromise?

currently Symfony provides 255 country codes (intl-data/region/en.json), with#31365 that is 252 even.

removing the aliases fromhttps://en.wikipedia.org/wiki/List_of_ISO_3166_country_codes gives us 249 codes.

I think we're very close already, and probably if we remove the remaining "islands" and "territories" we should be good.

But then, it becomes our responsibility to filter regions, which is probably something we want to do in the long run. Woudn't it "better" to renameCountryValidator instead?

which is probably something we want to do in the long run

want or "dont want"? We already filter "some" regions today, but now im unsure what the intend is :) "exclude invalid regions" (what's an invalid region?) vs. "exclude invalid countries" (we know that's ISO-3166).

If we dont filter "regions" at all the list effectively becomeshttps://github.com/unicode-org/icu/blob/master/icu4c/source/data/region/en.txt ... so001 (World) is a valid region?

from a domain perspective i value "countries" more than "regions", thus ISO-3166.

Sorry, I means "don't want" here :)

I do understand that we need to filter some entries that are not countries (World). But filtering countries is way more complex as Javier noted. I think it's not Symfony's responsibility to make choices here. So, sticking to ISO-3166 is an option, keeping regions is another one. I must admit that I don't have any preference here but I recognize that we need to make an informed decision here.

I'd vote for sticking with ISO-3166 for now, thus "countries" and filter what's not an ISO-3166 code

seehttps://www.diffchecker.com/OJeLGEYU it's doable.

the process is also easy IMHO, whenever we apply an ICU update we should verify only ISO-3166 country codes are added

@symfony/deciders We need your point of view here. Thank you.

in#31365 i've made the list iso compliant for comparison

if anything ... i'd say it was really close to that already, compared to "includes any region" :D

also there's#18613 which truly ask for an ISO list of countries

I'm for "Country" - the examples you gave Javier relate all to some geographical configuration. But if I want to create an "Address" field, I really want a "Country" label before entering "France". I think that's what this is for most of the time.

Sticking to ISO-3166 seems fine and then it's technically countries. If someone wants to use a more generic term like "region" they can just change the label. But for most cases it's for adress fields as Nicolas said, and there only countries make sense because you cannot send a postcard to the EU or EZ. Or can you? Would be funny to try out.

TLDR;

• ICU provides regions in form of continents, groups, "countries", etc. -http://icu-project.org/apiref/icu4c/classicu_1_1Region.html#details
• in practice the ICU region 2-letter codes form the ISO list already -[Intl] Rename Regions to Countries #31350 (comment)

Fromhttps://en.wikipedia.org/wiki/ISO_3166-1_alpha-2#Decoding_table

• ICU excludes "Deleted", "Unassigned", etc. already
• SF additionally should exclude "User-assigned" and "Exceptionally reserved"

After#31365 SF will provide the "officially assigned ISO 3166-1 alpha-2 codes"; those we call "Countries", hence this PR.

Thus, per ISO standards; e.g. Antarctica and Hong Honk are countries. Whether we can send a postcard to some address here is vague. For Hong Kong i believe yes, for Antarctica i believe not (yet there's a post office 😂https://theplanetd.com/a-post-office-in-antarctica-you-betcha/)

Im not aware of why Antarctica (a continent) is assigned an official country code, i guess it's some practical reason (as it's a continent without countries).

Im not aware of why Antarctica (a continent) is assigned an official country code, i guess it's some practical reason (as it's a continent without countries).

Actually, Antartica is split in pieces that rattached to existing countries:https://en.wikipedia.org/wiki/Antarctica#Antarctic_territories