Long lines in new ScriptExtensions.txt cause the space after # to disappear

[roozbehp]

Instead of # Po MIDDLE DOT, the first data line in https://github.com/unicode-org/unicodetools/blob/main/unicodetools/data/ucd/dev/ScriptExtensions.txt reads #Po MIDDLE DOT. I see no reason for this space to be dropped if the data is too long, while it's kept for lines with shorter data.

[eggrobin]

I had noticed this. It is weird, but it is consistent with what we do elsewhere, note in DerivedNormalizationProps.txt

FDF1          ; NFKC_CF; 0642 0644 06D2 # Lo       ARABIC LIGATURE QALA USED AS KORANIC STOP SIGN ISOLATED FORM
FDF2          ; NFKC_CF; 0627 0644 0644 0647 #Lo   ARABIC LIGATURE ALLAH ISOLATED FORM
FDF3          ; NFKC_CF; 0627 0643 0628 0631 #Lo   ARABIC LIGATURE AKBAR ISOLATED FORM
FDF4          ; NFKC_CF; 0645 062D 0645 062F #Lo   ARABIC LIGATURE MOHAMMAD ISOLATED FORM
FDF5          ; NFKC_CF; 0635 0644 0639 0645 #Lo   ARABIC LIGATURE SALAM ISOLATED FORM
FDF6          ; NFKC_CF; 0631 0633 0648 0644 #Lo   ARABIC LIGATURE RASOUL ISOLATED FORM
FDF7          ; NFKC_CF; 0639 0644 064A 0647 #Lo   ARABIC LIGATURE ALAYHE ISOLATED FORM
FDF8          ; NFKC_CF; 0648 0633 0644 0645 #Lo   ARABIC LIGATURE WASALLAM ISOLATED FORM
FDF9          ; NFKC_CF; 0635 0644 0649 # Lo       ARABIC LIGATURE SALLA ISOLATED FORM

and it seems to be intentional, see this comment:

unicodetools/unicodetools/src/main/java/org/unicode/props/BagFormatter.java

Lines 519 to 523 in 6f0c77d

	// old:
	// 0009..000D ; White_Space # Cc [5] <control-0009>..<control-000D>
	// new
	// 0009..000D ; White_Space #Cc [5] <control>..<control>
	tabber.add((mergeRanges ? 14 : 6) + minSpacesBeforeSemicolon, Tabber.LEFT);

I have no idea what the intention is though. The commit that added that comment is unicode-org/icu@cd418af, its message is not particularly illuminating, and neither is ICU-6106. @macchiati, do you remember what you were thinking 16 years ago?

[markusicu]

Let's stop doing this