Module talk:pi-decl/noun/testcases

Analysis of 39 errors of 29 September 2018
, : Today's testing reported 39 errors. 17 of them relate to 3 issues with the choice of round AA v. tall AA; the third of them, Burmese 'vo', has an obvious solution. 6 of them appear to be a grammar error (copy, paste and forget to change?) in the Thai module. The remaining 16 are due to an error (or just possibly, a lack of robustness) in the transliteration module. The errors are discussed in more detail below. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)

Errors 1 to 7
The transliteration module and the noun declension module are inconsistent as to the Tai Tham form of the sequence -ṅgā-, which occurs in the ablative singular and much of the plural of aṅga. The transliteration module renders the plural as ᩋᩘᨣᩣᨶᩥ, while the declension module renders it as ᩋᩘᨣᩤᨶᩥ. Investigation of practice across the four nations is indicated, as this combination may well be subject to special rules within each system. Maha Sena's Lao textbook will mandate ᩋᩘᨣᩣᨶᩥ, while the Tai Khün practice of writing mai kang lai as a narrow mark in the gap between and above the surrounding consonants leads me to expect them to use ᩋᩘᨣᩤᨶᩥ. There may also be some usage of plural ᩋᨦ᩠ᨣᩣᨶᩥ together with, for example, genitive/dative singular ᩋᩘᨣᩔ. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)
 * This has been solved for now. The Maha Sena Lao practice has been categorised as aberrant even in this case by  tweaking Module:pi-Latn-translit so it transliterates the sequence with tall AA.  RichardW57 (talk) 13:27, 30 September 2018 (UTC)

Errors 8 to 14
The transliteration module and the noun declension module are inconsistent as to the Burmese form of the sequence -ṅgā-, which occurs in the ablative singular and much of the plural of. The transliteration module renders the plural as အင်္ဂါနိ, while the declension module renders it as အင်္ဂာနိ. Investigation of Burmese practice is needed. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)

A Google search for အင်္ဂါနိ gets 499 hits, mostly, if not all, Pali, and we already have the Burmese derivative on Wiktionary. A search for အင်္ဂာနိ finds nothing. I'm going to treat this as a bug in Module:pi-decl/noun and tweak it so that the plural comes out as အင်္ဂါနိ. - RichardW57 (talk) 21:27, 29 September 2018 (UTC)

Errors 33 to 35
Roman script Nom/Voc/Acc bandhavo of is transliterated to Burmese script ဗန္ဓဝော, whereas it should be transliterated to ဗန္ဓဝေါ. This is a fault with Module:pi-Latn-translit. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)

Errors 15, 32, and 36 to 39
The Thai script module has instrumental singular of masculine and neuter i-class in -iyā; other scripts' modules have -inā. The Thai script module consistently has instrumental singular of masculine and neuter u- and ū-class in -uyā; other scripts' modules have -unā.

This should be treated as an issue with Module:pi-decl/noun/Thai. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)

Errors 16 to 31
The problem here lies with Module:pi-Latn-translit. It does not recognise the digraph in 'dh' in the sequence 'dhy', and the 'expected' column is therefore showing it as a conjunct of 3 letters d-h-y. If does indeed follows the 'regular' paradigm, it shows up in the oblique singular, though with the current tables this is restricted to the ablative and locative singulars. - RichardW57 (talk) 20:08, 29 September 2018 (UTC)

Test Too Big
It seems that the consistency test is too big. The testcase module page is being placed in Category:Pages where template include size is exceeded; one currently has to view the test results at Module:pi-decl/noun/testcases/documentation. I will attempt to solve this by moving the tests of Roman script against each other script out to that other script. This will not be so good for identifying common problems. RichardW57 (talk) 17:22, 30 September 2018 (UTC)