CLDR-19157 Restrict era codes to 8 characters #5283

robertbastian · 2026-01-12T13:56:18Z

CLDR-19157

This PR completes the ticket.

ALLOW_MANY_COMMITS=true

jira-pull-request-webhook · 2026-01-14T17:27:11Z

Notice: the branch changed across the force-push!

tools/cldr-code/src/test/java/org/unicode/cldr/util/TestSupplementalDataInfo.java is now changed in the branch

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

macchiati

I don't see a change to the spec that limits the length to 8 (for 48 and beyond)

I think you said that 48 got rid of the ones that were longer, if different then adjust.

robertbastian · 2026-01-14T23:37:23Z

what exactly do you mean?

srl295 · 2026-01-15T00:05:07Z

tools/cldr-code/src/test/java/org/unicode/cldr/util/TestSupplementalDataInfo.java

+            final CalendarData c = cal.get(calType);
+            for (final Integer n : c) {
+                if (c.get(n).getCode() != null) {
+                    assertTrue(c.get(n).getCode().length() <= 8);


Suggested change

assertTrue(c.get(n).getCode().length() <= 8); final String code = c.get(n).getCode(); assertTrue(code.length() <= 8, () -> "code " + code + " > 8");

otherwise it will show "assertion failed: false", but with this it will fail with "long era code: abcdefghij > 8"

srl295

Looks good, with a suggestion for a better err message

macchiati · 2026-01-15T02:05:14Z

docs/ldml/tr35-dates.md

-Each `era` element has a `code` attribute and optional `aliases` attributes that define invariant strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below).
-The `code` is unique within the calendar, and the `aliases` are space-separated identifiers, each also unique within the calendar.
+Each `era` element has a `code` attribute and optional `aliases` attributes that define stable strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below).
+The `code` defines the primary identifier for the era, and `aliases` are space-separated additional identifiers. All identifiers are valid BCP-47 subtags, i.e. 2-8 ASCII alphanumeric characters.


This says that each subtag is 2-8 characters. There_ is no statement that the total length of the code (including all subtags + separators) is limited to 8 characters, as per:

Agreed in Design WG that additional limit of 8 characters can be added to the spec, with a unit test for all versions ≥ 48.

It says that each identifier is a BCP-47 subtag. It also says that the code is a single identifier.

This was a limit of 8 characters on the code attribute.

macchiati · 2026-01-15T15:39:21Z

That isn't what we agreed on.

…

On Thu, Jan 15, 2026, 02:31 Robert Bastian ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In docs/ldml/tr35-dates.md <#5283 (comment)>: > @@ -1098,8 +1098,8 @@ For example, the following means that the two eras from calendar "gregorian" sho </calendar> ``` -Each `era` element has a `code` attribute and optional `aliases` attributes that define invariant strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below). -The `code` is unique within the calendar, and the `aliases` are space-separated identifiers, each also unique within the calendar. +Each `era` element has a `code` attribute and optional `aliases` attributes that define stable strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below). +The `code` defines the primary identifier for the era, and `aliases` are space-separated additional identifiers. All identifiers are valid BCP-47 subtags, i.e. 2-8 ASCII alphanumeric characters. It says that each identifier is a BCP-47 subtag. It also says that the code is a single identifier. — Reply to this email directly, view it on GitHub <#5283 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACJLEMAMQG72FGS35HAHCCT4G5UAPAVCNFSM6AAAAACRNOYWU6VHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTMNRUHE2DMMJQGI> . You are receiving this because your review was requested.Message ID: ***@***.***>

robertbastian · 2026-01-15T16:04:21Z

Then we need to discuss this again because that's what I understood.

macchiati · 2026-01-15T18:52:55Z

docs/ldml/tr35-dates.md

-Each `era` element has a `code` attribute and optional `aliases` attributes that define invariant strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below).
-The `code` is unique within the calendar, and the `aliases` are space-separated identifiers, each also unique within the calendar.
+Each `era` element has a `code` attribute and optional `aliases` attributes that define stable strings for identifying the eras. These are more mnemonic than the `type` identifiers (see below).
+The `code` defines the primary identifier for the era, and `aliases` are space-separated additional identifiers. All identifiers are valid BCP-47 subtags, i.e. 2-8 ASCII alphanumeric characters.


This was a limit of 8 characters on the code attribute.

macchiati · 2026-01-15T18:55:54Z

That is, it would be possible to have code='abc-xyz', but not code='abcd-wxyz' (since the latter would exceed 8 ASCII characters).

github-actions bot assigned robertbastian Jan 12, 2026

robertbastian force-pushed the eracode branch from c2cef2b to 0cbe2bf Compare January 12, 2026 14:04