CLDR-18464 Add a Before Hijrah era to the Islamic calendars

[sffc]

CLDR-18464

• This PR completes the ticket.

ALLOW_MANY_COMMITS=true

CC @roozbehp @pedberg-icu @AEApple

[sffc]

I gave it the code "islamic-inverse" because I want to land the code updates, #4519, atomically once we have agreement on them.

[sffc]

I gave it the type -1 because type 0 is already used for the AH era. Please let me know if I should instead:

1. Use a positive number like 1 (and leave AH as 0)
2. Change BH to 0, AH to 1, and update all the XML files that currently have translations for AH at type 0 to instead be at type 1

[pedberg-icu]

@sffc Usually this number is used as an array index in ICU. I guess I would be in favor of your option #2 above. We would need to add a migration note for ICU (and for CLDR), and probably update some ICU code. I think we allow years to be negative, but I do not think we allow eras to be.

[roozbehp]

Generally looks good to me. I would also prefer the "Change BH to 0, AH to 1" option, as it seems cleanest.

[sffc]

This means changing all the XML files, which I'm willing to do and can probably automate, but to make sure, before I proceed, is it sufficient for me to change something like

					<eraAbbr>
						<era type="0">যুগ</era>
					</eraAbbr>

to

					<eraAbbr>
						<era type="1">যুগ</era>
					</eraAbbr>

without adding structure for type="0"? It will inherit automatically, right?

[pedberg-icu]

This means changing all the XML files, which I'm willing to do and can probably automate...It will inherit automatically, right?

As far as I know, yes.

[sffc]

Does this look right?

Here's how I generated the diff:

$ xmlstarlet ed --inplace -P -u "/ldml/dates/calendars/calendar[@type='islamic']/eras/*/era[@type=0]/@type" -v 1 *.xml

That updated all the XML files, but it also resulted in some other unwanted diffs to the XML. I undid those by manually touching up the diff file to remove diffs on lines containing &quot. I then went through the remaining changes and verified that they were touching only the era IDs.

To verify that I did everything, correct, I then ran:

$ xmlstarlet sel -t -v "/ldml/dates/calendars/calendar[@type='islamic']/eras/*/era[@type=0]" *.xml
Warning: program compiled against libxml 212 using older 209
Before Hijrah
BH
BH
$

So it found the three BH entries I put in and nothing else.

[sffc]

The other calendars don't have data. Only islamic.

$ xmlstarlet sel -t -v "/ldml/dates/calendars/calendar[@type='islamic-civil']/eras/*/era[@type=0]" *.xml
$ xmlstarlet sel -t -v "/ldml/dates/calendars/calendar[@type='islamic-rgsa']/eras/*/era[@type=0]" *.xml
$ xmlstarlet sel -t -v "/ldml/dates/calendars/calendar[@type='islamic-umalqura']/eras/*/era[@type=0]" *.xml
$ xmlstarlet sel -t -v "/ldml/dates/calendars/calendar[@type='islamic-tbla']/eras/*/era[@type=0]" *.xml
$ 

[srl295]

needs commit fix

[sffc]

@srl295, do you prefer me to merge the PR in this shape (AH becomes index 1, BH inserted at index 0) or would you prefer keeping AH at index 0 and adding BH at index 1 as @FrankYFTang suggested?

[pedberg-icu]

@srl295, do you prefer me to merge the PR in this shape (AH becomes index 1, BH inserted at index 0) or would you prefer keeping AH at index 0 and adding BH at index 1 as @FrankYFTang suggested?

@sffc I am not @srl295 but I now prefer keeping AH at index 0 and adding BH at index 1

[sffc]

OK please re-review.

[srl295]

The overall PR is straightforward to review now, with changes in other locales undone

[srl295]

Do we need a followup issue to make sure CLDR spec clarifies this question so that it doesn't need to get discussed again? Variants and/or lack thereof?

[jira-pull-request-webhook]

Hooray! The files in the branch are the same across the force-push. 😃

~ Your Friendly Jira-GitHub PR Checker Bot

[sffc]

This is ready to merge; should I Squash And Merge or Rebase And Merge?

I'd like to use CLDR-18369 for all spec updates on this topic.

[pedberg-icu]

This is ready to merge; should I Squash And Merge or Rebase And Merge?

We usually Squash and merge except for special circumstances such as cherry-picking maint branch changes to main. But since this only has 1 commit does it make a difference? Does squash gives a different commit hash?

[pedberg-icu]

This causes an IndexOutOfBoundsException at org.unicode.cldr.test.ExampleGenerator.handleEras(ExampleGenerator.java:3158), because there is a reference to a hard-coded CALENDAR_ERAS HashMap maps "islamic" to a sample date for only a single era, need to add another entry for the new islamic era. I will make a PR to do that.