The new version 2.0 of the (Mozilla) Firefox web browser supports spell checking in web forms. This is a new feature with some obvious deficiencies that might be improved in coming versions.
Has there been any discussion, proposal or attempts to adopt Firefox' spell checker to Wikipedia's needs? Currently, wiki syntax is not recognized, and the language needs to be set manually. If I go to nl.wikipedia.org and click edit, I want the browser to automatically select the Dutch spelling dictionary. And there is no need to apply spell checking to interwiki links, external URLs, template names or template parameter names. Perhaps the website needs to communicate this to the browser by some tags to the HTML, e.g. <textarea lang=nl syntax=mediawiki>?
Currently the language support in Firefox is only a spell checker. As a comparison, OpenOffice also offers automatic corrections (if I type "abotu", this can automatically be corrected to "about"), hyphenation, and a thesaurus. I have no idea what is planned, but I guess these features will eventually creep into Firefox too.
One option offered by the interaction between browser and website is that the spelling dictionary could be maintained and updated online. For example, if I spell "abotu", a word not found in the browser's built-in dictionary, the browser could send a request to Wiktionary (-Z) for advice on that word. This option is not so obvious to developers of OpenOffice as it should be to developers of Firefox. This opens up a new can of legal worms: Should Wiktionary offer spelling help only for documents that are released under a free license?
Lars Aronsson wrote:
This opens up a new can of legal worms: Should Wiktionary offer spelling help only for documents that are released under a free license?
I believe this would count as "use" rather than as "creation of a derivative work". Copyleft licenses, as a rule, do not in any way restrict the way a copy of the work may be used by its owner -- indeed, I believe at least the GPL (not sure about the GFDL) even explicitly forbids creators of derivative works from imposing such restrictions.
Of course, if such a feature were to be implemented, we could certainly choose to restrict the use of our server resources in any way we'd like. But someone else could always download a database dump and use that as they will.
Lars Aronsson schreef:
The new version 2.0 of the (Mozilla) Firefox web browser supports spell checking in web forms. This is a new feature with some obvious deficiencies that might be improved in coming versions.
Has there been any discussion, proposal or attempts to adopt Firefox' spell checker to Wikipedia's needs? Currently, wiki syntax is not recognized, and the language needs to be set manually. If I go to nl.wikipedia.org and click edit, I want the browser to automatically select the Dutch spelling dictionary. And there is no need to apply spell checking to interwiki links, external URLs, template names or template parameter names. Perhaps the website needs to communicate this to the browser by some tags to the HTML, e.g. <textarea lang=nl syntax=mediawiki>?
Currently the language support in Firefox is only a spell checker. As a comparison, OpenOffice also offers automatic corrections (if I type "abotu", this can automatically be corrected to "about"), hyphenation, and a thesaurus. I have no idea what is planned, but I guess these features will eventually creep into Firefox too.
One option offered by the interaction between browser and website is that the spelling dictionary could be maintained and updated online. For example, if I spell "abotu", a word not found in the browser's built-in dictionary, the browser could send a request to Wiktionary (-Z) for advice on that word. This option is not so obvious to developers of OpenOffice as it should be to developers of Firefox. This opens up a new can of legal worms: Should Wiktionary offer spelling help only for documents that are released under a free license?
Hoi, The suggestion that a spell checker created from open content could only operate on Open/Free content is absolutely brain dead. The point why you want to build a spell checker is to provide better support to the people who are in need of a spell checker. You do not open a can of legal worms, spell checkers can operate on any content, you do not create a derivative work in that way.
What is needed in the creation of content and in the use of the right spell checker is for content to be marked correctly as to the language it is. According to a presentation of Google only 15% of the content of the web has an indication as to what language is used. Much of these 15% are tagged incorrectly. Thanks, GerardM
We already communicate to the browser that nl.wikipedia.org is in dutch, I haven't looked at the firefox 2.0 spell checker but I suspect they don't do any language detection at all since:
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="nl" lang="nl" dir="ltr">
is *the* standard way to say that a web page (and by extension any text areas on it that don't have their own lang and xml:lang attributes) is written in dutch.
With regards to the legal issues Gerard is right, there are none.
On 12/13/06, Lars Aronsson lars@aronsson.se wrote:
And there is no need to apply spell checking to interwiki links, external URLs, template names or template parameter names. Perhaps the website needs to communicate this to the browser by some tags to the HTML, e.g. <textarea lang=nl syntax=mediawiki>?
What about syntax? Would we want it to ignore <nowkii> or attempt to fix it?
<drools thinking about how cool intellisense for Wikipedia would be...>
Steve
wikitech-l@lists.wikimedia.org