Bug 38409 - regression in hu spellchecking (3-ban, %-kal)
Summary: regression in hu spellchecking (3-ban, %-kal)
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Linguistic (show other bugs)
(earliest affected)
3.4.0 release
Hardware: Other All
: medium normal
Assignee: Not Assigned
Keywords: regression
Depends on:
Reported: 2011-06-17 07:24 UTC by Andras Timar
Modified: 2011-11-14 13:12 UTC (History)
1 user (show)

See Also:
Crash report or crash signature:

debugging patch to show what words hit the spell-checking and the return value (832 bytes, patch)
2011-11-14 05:18 UTC, Caolán McNamara
demo .odt (8.87 KB, application/vnd.oasis.opendocument.text)
2011-11-14 06:17 UTC, Caolán McNamara
screenshot (2.76 KB, image/png)
2011-11-14 06:19 UTC, Caolán McNamara

Note You need to log in before you can comment on or make changes to this bug.
Description Andras Timar 2011-06-17 07:24:12 UTC
In LibreOffice 3.3 and previous versions the Hungarian spellchecker accepted words beginning with numbers or other non-letter character + dash + suffix.

For example "3-ban" and "%-kal" were correct, while "3-ben" and "%-al" was wrong.

In 3.4 and above the correct forms are no longer accepted. When I enter "3-ban", LibreOffice underlines with red the suffix "ban" and also when I enter "%-kal", it underlines with red the suffix "kal". 

The breakiterator rules were not changed (i18npool\source\breakiterator\data\dict_word_hu.txt) but now those are ignored.
Comment 1 Andras Timar 2011-06-17 07:25:17 UTC
@Caolán: do you have an idea?
Comment 2 Caolán McNamara 2011-11-14 05:18:36 UTC
Created attachment 53524 [details]
debugging patch to show what words hit the spell-checking and the return value
Comment 3 Caolán McNamara 2011-11-14 06:17:52 UTC
Created attachment 53527 [details]
demo .odt
Comment 4 Caolán McNamara 2011-11-14 06:19:47 UTC
Created attachment 53528 [details]

So, this seems to work fine for me. The spell-checker itself marks the correct words as ok and the incorrect words as not ok. Higher levels decide to not mark anything starting with a number as incorrect.

Is there still a bug here ?
Comment 5 Andras Timar 2011-11-14 13:12:39 UTC
The bug was originally reported for 3.4.0. It somehow got fixed between 3.4.0 and now. In fact I cannot reproduce it with 3.4.2. I did not check 3.4.1. It is also working well in master. Thanks for looking into it.