Qt source code doesn't contain numerus info for the Lojban language

[vedgy]

When I run the command lupdate goldendict.pro, the following warning appears in the output:

File /path/to/goldendict/locale/jb_JB.ts won't be updated: it contains translation but the target language is not recognized

Neither renaming the file to e.g. jbo_IN.ts nor setting the language attribute of the TS element to say "jbo" allows to update this .ts file.

The issue was introduced in Qt 5.13 and cherry-picked into Qt 5.12.4: https://codereview.qt-project.org/c/qt/qttools/+/255003. The relevant Qt Linguist code: https://codebrowser.dev/qt5/qttools/src/linguist/lupdate/main.cpp.html#371. The .ts file is not updated, because the Lojban language is absent from numerusTable. The Lojban language is also absent from Qt6's numerusTable.

A workaround is to set the language attribute of the TS element to a supported language code, e.g. "en", but then plural translations could be broken. Does GoldenDict even contain plural translations?

@Abs62, do you update the ts files using an older version of lupdate unaffected by this issue?

@robintown, do you know which numerusTable entry is suitable for the Lojban language? We could work around the issue by setting the wrong language attribute value in jb_JB.ts permanently or only before running lupdate (then revert to jb or jbo). But the language information should also be contributed to Qt in order to prevent wrong translation info or a perpetual maintenance burden.

[Abs62]

do you update the ts files using an older version of lupdate unaffected by this issue?

I use lupdate from Qt4 as a rule.

[robintown]

I believe the right table entry would be the first one (Japanese style, nplurals=1; plural=0;). However I added that translation just for fun as a project a few years ago and I don't believe anyone's been using it since then, so if it's making life difficult for you all, don't hesitate to remove it.

[vedgy]

I believe the right table entry would be the first one (Japanese style, nplurals=1; plural=0;).

If I understand correctly nplurals=1 means no distinction between between singular and plural form of a word. But Lojban grammar appears to append s to the plural form of a word. This leaves a few choices depending on how "zero" multiplicity is treated in Lojban: same as "two" (English-style), same as "one", or a category of its own.

However I added that translation just for fun as a project a few years ago and I don't believe anyone's been using it since then, so if it's making life difficult for you all, don't hesitate to remove it.

If this translation is not very important, I suppose we'll eventually set the wrong (but best-fit) language attribute value in jb_JB.ts permanently to avoid any maintenance overhead.

[robintown]

No, words in Lojban do not actually have a separate plural form. The page you found is misleading, it appears to be from an unfinished series of lessons that were copied directly from a Spanish course, with only the first two pages in the series having been changed to feature Lojban rather than Spanish. For example "le kabri" would be translated as either "the cup" or "the cups" depending on context.

[vedgy]

Thank you for the explanation. #1651 sets the language attribute value to "ja" (Japanese).