Double hyphens in some compounds (Bugzilla Bug 633)

[albbas]

This issue was created automatically with bugzilla2github

Bugzilla Bug 633

Date: 2008-01-25T13:10:21+01:00 From: Sjur Nørstebø Moshagen <> To: Tomi Pieski <> CC: thomas.omma

Last updated: 2008-03-05T12:35:05+01:00

[albbas]

Comment 2504

Date: 2008-01-25 13:10:21 +0100 From: Sjur Nørstebø Moshagen <>

As shown in the new test reports:

http://www.divvun.no/doc/proof/spelling/testing/hy-wordtypes-pl-forrest-sme-20080125.html http://www.divvun.no/doc/proof/spelling/testing/hy-wordtypes-pl-forrest-smj-20080125.html

there are several instances of double hyphens in compounds, all related to the first part requiring a hard hyphen as part of the compound.

Since the latest speller/hyphenation lexicons are older than Tomi's changes to avoid this in our lexicons, we don't know whether that change will help. But I do suspect that this problem is also related to Polderland's code, as some or all of these compounds are dynamic compounds, where the soft hyphen is most likely inserted by the Polderland code/hyphenator.

[albbas]

Comment 2507

Date: 2008-01-25 15:22:47 +0100 From: Sjur Nørstebø Moshagen <>

The links are changed. They are now:

http://www.divvun.no/doc/proof/hyph/testing/hy-wordtypes-pl-forrest-sme-20080125.html http://www.divvun.no/doc/proof/hyph/testing/hy-wordtypes-pl-forrest-smj-20080125.html

[albbas]

Comment 2514

Date: 2008-01-29 11:32:47 +0100 From: Tomi Pieski <>

Most of these entries that get double hyphens from hyphenator are corrected properly in speller.

[albbas]

Comment 2516

Date: 2008-01-29 12:45:37 +0100 From: Sjur Nørstebø Moshagen <>

The problem words indicate a bug in the Polderland library. Changing component accordingly.

[albbas]

Comment 2518

Date: 2008-01-29 14:00:41 +0100 From: Sjur Nørstebø Moshagen <>

To Comment #2:

In Word, none of:

AP--rávvagat ANC--reahccut 10--jahkásačča Davimušaláš--geađggit Koskivuori--plánenreaiddut

get flagged, and thus none are corrected.

This is using:

Davvisámi, version 1.0.1, 2008-01-28

There is actually a discrepancy between the command line and Word versions in this respect: the command line tool flags these misspellings, whereas the Word speller does not.

Feedback from Polderland would be welcome.

[albbas]

Comment 2588

Date: 2008-02-08 10:04:45 +0100 From: Sjur Nørstebø Moshagen <>

Comment from Polderland:

(feature) double hyphenation in compounds ("ANC--reahccut") True, but feature, not bug: the cmdline tool lists "--" for hard hyphen and "-" for soft hyphen

[albbas]

Comment 2589

Date: 2008-02-08 10:07:16 +0100 From: Sjur Nørstebø Moshagen <>

Thanks for the feedback.

Reducing priority and severity, as well as removing Polderland from the Cc list. We need to update the test bench to properly account for this (useful) feature.

[albbas]

Comment 2594

Date: 2008-02-08 10:51:43 +0100 From: Sjur Nørstebø Moshagen <>

Re-adding Peter just in case.

When the test bench is fixed, we still have one test case that doesn't pass the test:

Koskivuori-plánenreaiddut

The word is accepted by the speller (Davvisámi, version 1.0.1, 2008-01-31), which means that there are either wrong hyphenation points in each of the parts of this compound (most likely), or an undetected hyphenation bug in the Polderland code.

Tomi, could you please give the PLX entries for:

Koskivuori Koskivuori- plánenreaiddut

?

[albbas]

Comment 2618

Date: 2008-03-05 10:38:05 +0100 From: Sjur Nørstebø Moshagen <>

This is a non-issue. It turns out that the double hyphen is a feature of the command-line hyphenator, designating a hard hyphen (just as in the PLX entries).

What is needed is that we update the test bench to consider the difference between soft and hard hyphens.

[albbas]

Comment 2622

Date: 2008-03-05 12:35:05 +0100 From: Sjur Nørstebø Moshagen <>

The test bench is now updated to correctly handle this case.

gt/script/hyphen-testres.pl new revision: 1.17

Tested and found working.

Case closed.