216267 - Spellchecker suggest splitting up correctly typed compounds

Status: RESOLVED WORKSFORME

(Core :: Spelling checker, defect)

Description

[Toralf Lund]

User-Agent:       Mozilla/5.0 Galeon/1.2.6 (X11; Linux i686; U;) Gecko/20020913
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003

In most germanic languages, a compound should always be written as one word.
E.g. the "bug database" might be translated into "feildatabase" in Norwegian,
but "feil database" would not be correct - in fact this means something else;
"wrong database" or similar. Unfortunately, the Mozilla spellchecker won't
understand this; it will suggest replacing "feildatabase" with "feil database".

The strange part here is that OpenOffice does not make this mistake when using
the same dictionary - as its implementation of MySpell actually has special
support for compounds, but that part seems to be missing from the Mozilla
incarnation.



Reproducible: Always

Steps to Reproduce:
1. Download Norwegian dictionary from OpenOffice dictionary pages, and install
in "components/myspell
2. Type the word "feildatabase" e.g. in mail composer.
3. Spell check with language set to Norwegian (language/country name may not be
displayed correctly; select no-NO if it isn't.)  
4. Repeat 2 and 3 in OpenOffice.org

Actual Results:  
For Mozilla:
Dialoge for incorrect spelling opened for the word, with "feil database"
suggested as replacement

For OpenOffice:
The spellchecker accepts the word

Expected Results:  
Word accepted by spellchecker.

This does not happen for all compounds, as they may be listed in the dictionary
themselves. However, including all combinations of words that may be written
together is not realistic for German or the Nordic languages, unlike e.g. for
English where only a limited number of well-know compounds are written that way.

The example uses Norwegian, but the problem can probably be reproduced e.g. for
German as well.

The Mozilla version indicated here is fairly old, but I get the same problem
with the most recent snapshot of Thunderbird.

See also http://folk.uio.no/tfredvik/amo/

Comment 1

[Toralf Lund]

The problem is probably related to the fact that parts of current OOo MySpell is
missing from => add dependency to Bug 240600. (Or is this even a duplicate?)

Comment 2

[Michiel van Leeuwen (email: mvl+moz@)]

*** Bug 263983 has been marked as a duplicate of this bug. ***

Comment 3

[u60234]

With a Swedish dictionary in the latest trunk builds of Seamonkey and
Thunderbird, the spellchecker accepts compound words like "adressbok" and
"konsulttjänster". So bug 240600 seems to have fixed this.

Comment 4

[Fredrik Toftered]

I downloaded the latest 1.8b and a swedish dictionary from
http://www.mozilla.org/products/thunderbird/dictionaries.html

Then I tested with the following words:

Adressbok konsulttjänst vinprovning bilprovning besiktningsman

Mozilla suggested every word to be splitted up.

Hasee has probably previously added adressbok and konsulttjänst to his profile's
.dat file.

Fredrik 

Comment 5

[u60234]

Hmm, the Swedish dictionary on
http://www.mozilla.org/products/thunderbird/dictionaries.html
is an old one and does not seem to have support for compounds.

I have been using the latest files from http://sv.speling.org/myspell.html and
they will happily accept words like "vinprovningskonsulttjänster".

http://dictionaries.mozdev.org/installation.html also have an up-to-date Swedish
dictionary.

Comment 6

[Fredrik Toftered]

You are right Hasse. The issue seems to be fixed. :-)

Fredrik

Comment 7

[Michiel van Leeuwen (email: mvl+moz@)]

ok