321742 - Wrong dictionary included in en-GB localisation

Status: NEW

(Thunderbird :: Installer, defect)

Description

[Steven Foster]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.8) Gecko/20051111 Firefox/1.5
Build Identifier: http://download.mozilla.org/?product=thunderbird-1.5rc2&os=win&lang=en-GB

The en-GB localised build of TB 1.5 RC2 comes with the en-US dictionary instead of the en-GB dictionary.


Reproducible: Always

Steps to Reproduce:
1. Download and install once more.
2.
3.

Actual Results:  
As described above.

Expected Results:  
Not required me to search for and download the en-GB dictionary.

I have read bug #310937 but that bug report does not sugest that the local dictionary is excluded altogether, hence I guess I'm filing a valid bug report.
If this is fixed by adding the en-GB dictionary alongside en-US as per bug #310937 then en-GB should be selected as the default. However, I'm sure 99% of Brits would like to see en-US taken out as being unnecessary for them.

Comment 1

[Adam Kennedy]

Confirmed in production 1.5 installer. Installs en-US, not en-GB.

And since, there's no en-AU localisation, you can also include pretty much all Australian also not needing or wanting en-US to be installed as well.

Comment 2

[Adam Kennedy]

While this isn't a dupe of bug #321742 it almost certainly relies on that whole enforced inclusion of American dictionaries problem to be fixed.

To really fix this bug, you want American not there at all. Having both of en-GB and en-US is a half-assed solution. It should be there at all.

Adding as bug #310937 as a tentative dep.

Comment 3

[Mark Tyndall]

The problem is, the en-GB dictionary that is available is licensed under the LGPL.  Due to mozilla.org policy (as I understand it), the official binaries do not include such material.

There are three ways to solve this problem.
1) Create your own dictionary, from scratch, and tri-license it;
2) "Launder" the existing dictionary, which is morally, if not legally, dubious and a non-trivial task to boot;
3) Get a change of policy to admit dictionaries as a(nother) special case.

None of the three is simple, though one is politically, rather than technically, difficult (sorry, Gerv).

cc'ing Gerv, who knows about licensing.

Comment 4

[Gervase Markham [:gerv]]

We intend to have a discussion about the dictionary issue at FOSDEM - are you coming?

en-GB may be a special case, because it may well be possible to make a free en-GB dictionary from the en-US one by combining it with a free list of the differences between en-GB and en-US, which you may be able to find on the web.

Gerv

Comment 5

[u63580]

Is it likely that a new en-GB dictionary is going to be needed?  If so, this seems rather a waste of effort, as the existing OO one is pretty reasonable (apart from the full-stops problem).  However, if a new one is needed, I assume there will be a round up of British (plus Australian?) volunteers.

Comment 6

[Adam Kennedy]

(In reply to comment #5)
> Is it likely that a new en-GB dictionary is going to be needed?  If so, this
> seems rather a waste of effort, as the existing OO one is pretty reasonable
> (apart from the full-stops problem).  However, if a new one is needed, I assume
> there will be a round up of British (plus Australian?) volunteers.

And I imagine you could include quite a bit of the rest of the Commonwealth in there too?

The full list would be England, Australia, New Zealand, South Africa, Canada, India and a few others that need the en_gb instead of the en_us as their default English dictionary.

Comment 7

[Gervase Markham [:gerv]]

> Is it likely that a new en-GB dictionary is going to be needed?

If the current one stays under the LGPL, and you want to include a dictionary directly in distributed builds, then yes.

Gerv

Comment 8

[u63580]

(In reply to comment #7)
> > Is it likely that a new en-GB dictionary is going to be needed?
> If the current one stays under the LGPL, and you want to include a dictionary
> directly in distributed builds, then yes.
> Gerv

Oh well.  Personally, I think it looks very poor not to have an en-GB dictionary in the en-GB version of TB (the same is of course true for any localisation).  So that means that a new dictionary will be needed.  Is anyone on the localisation team "in charge" of this, or is it to be decided?  I've no experience with hoe the dictionaries are actually produced, so I wonder how easy this is all going to be.

Comment 9

[Mark Tyndall]

[sorry, I didn't see many comments on this bug for some reason]

(In reply to comment #8)
> [...]  Is anyone
> on the localisation team "in charge" of this, or is it to be decided?  I've no
> experience with hoe the dictionaries are actually produced, so I wonder how
> easy this is all going to be.

There's no-one in charge of it, at the moment.

David Bartlett, who produces the LGPL dictionary, will know about the difficulty, but I'm not sure if he has a Bugzilla account.  His email address is in the LGPL dictionary's README.  I don't know how willing he'll be to duplicate his efforts, but you can always ask.  Note that the affix file is not based on earlier work, so he may be willing to re-license that - but I don't know how tightly coupled that is with the main .dic file.

Comment 10

[kuro68k]

I would be willing to take responsibility for the en-GB dictionary, and perhaps the entire localisation effort if required. Some pointers to the tools required for localising would be helpful.

It's worth noting that British English is not entirely the same as International English too. Also, in en-GB sometimes multiple spellings are acceptable, such as -ise and -ize words.

This is an important issue though. A lot of very common words are spelt differently in UK English, and if you spell them wrong you look like an idiot (e.g. check instead of cheque).

Comment 11

[Wayne Mery (:wsmwk)]

can someone check en-GB situation in beta 1 builds?
  http://www.mozillamessaging.com/en-US/thunderbird/early_releases/

Comment 12

[u63580]

The installer does not include a dictionary at all: I assume this is due to the same license issue that means the FF localisation doesn't have one included.

Comment 13

[Wayne Mery (:wsmwk)]

ah, right it would.

Paul, any luck contacting Bartlett?

Thunderbird v3 comes out in several months and I'm sure some people would welcome a shipping en-GB dictionary.

Comment 14

[Wayne Mery (:wsmwk)]

Marco, can you test?  What do you think?

Comment 15

[Marco A.G.Pinto [:marcoagpinto]]

(In reply to Wayne Mery (:wsmwk) from comment #14)
> Marco, can you test?  What do you think?

I don't understand what there is to test.

I have in TB 31 the en_GB forked by me which has over 10 000 new words compared to Bartlett (and whatever guys who vanished aeons ago).

If you wish, you can include my version with FF, TB and SM. I am here to help mankind. I am silly but I struggle hard.

I tried to contact the "REAL" guys years ago but I never received any reply, or the e-mail addresses no longer existed.

Comment 16

[Wayne Mery (:wsmwk)]

Does bug 1326038 comment 6 mean there should be no problem including a dictionary in our install?

Comment 17

[Magnus Melin [:mkmelin]]

No problem (except some technical work) in case the dictionary license is compatible. But of course there are complications when they all are not compatible, or probably missing for some language + some languages having more than one.

Comment 18

[:aceman]

I don't know more here. What dictionary do we bundle with en-GB version of TB? None? en-US as the report claims?
If so, do we have a en-GB dictionary available to be included? It seems Marco would give us permission to use his one.
I haven't found which licence his dictionary is under, so that should be clearly spelled out on the webpage and in the dictionary package.

Downloading Tb from https://www.thunderbird.net/en-US/thunderbird/all/ shows:
1. en-US TB contains en-US dictionary
2. en-GB TB contains NO dictionary

Comment 19

[Marco A.G.Pinto [:marcoagpinto]]

(In reply to :aceman from comment #18)
> If so, do we have a en-GB dictionary available to be included? It seems
> Marco would give us permission to use his one.
> I haven't found which licence his dictionary is under, so that should be
> clearly spelled out on the webpage and in the dictionary package.


@aceman:
The licence is LGPL:
https://github.com/marcoagpinto/aoo-mozilla-en-dict

For ages that I have been suggesting the use of my speller.

Will it now be officially used?

Also, can you guys do like Calendar which appears as a normal add-on so that users don't complain:
"I downloaded Marco's speller from the add-ons site and it only has two or three new words compared to the one that ships with FF, TB and SM"?

This way it would appear with the version number in the "dictionaries" tab and it would automatically update every month.

Thank you!

Kind regards from the cola demon,

Comment 20

[najoll]

Is this why my British dictionary disappears after every update? That seems a serious bug - it will affect many users. At the least - given the difficulties mentioned above - some sort of prompt or notice could be displayed. It seems this bug - a serious one, as I say, or at least a recurring, widespread inconvenience - was reported over *a decade* ago. That could win an award (that one would not want to win).