Inline Spell-check Problems, Comments, Tips

Inline Spell-check Problems, Comments, Tips

This is enabled by default. You can turn it off in Opera Config Spellchecker

Direct links to Opera:Config items are still broken. Go to User Prefs and scroll down from there.

Originally posted by Heureka:

I have never tried Aspell or Ospell, so I can't comment on their quality. Note however that there was a spelling reform for Dutch in 2005 and the Dutch Aspell dictionary hasn't been updated.

I have done some tests comparing the Dutch Opera spell checker/spell-checker/spellchecker to the one from OpenOffice.org (and Firefox). In general Opera's results are inferior: OpenOffice.org and Firefox find more spelling errors or include the correct proposal that Opera doesn't offer.
Both are using the same dictionaries (and aff files), so I suppose it's the 'spelling engine' that causes these problems.

Here are some (Hunspell) problems:

1. the hyphen: hyphenated words aren't checked as a whole but on their parts. Sometimes this method is good (e.g. the flight London-Manchester, London-Paris, London-Sydney ...; British-Australian relations, British-French relations, ...) but in many cases it's not. Let me give you an example from Dutch: Kapellen and Kapelle-op-den-Bos are both Flemish town names included in the Dutch dictionary. If you type Kapellen-op-den-Bos the error will not be detected because Kapellen and op and den and bos (or Bos) are all valid Dutch words, but the combination of these four words is completely wrong.

2. upper case/lower case: the Opera spell checker takes for granted that every word can be capitalized/capitalised, even in the middle of a sentence. This leads to correct results in (American) headings and (book or film) titles, but often fails.
In a Dutch text I found Kerst en nieuwjaar (the correct spelling is kerst en Nieuwjaar, meaning Xmas and New Year): nieuwjaar was underlined as an error, but unfortunately Kerst was not. If you write Kapelle-Op-Den-Bos instead of Kapelle-op-den-Bos, it will pass. Maybe Stratford-upon-Avon and Stratford-Upon-Avon are both acceptable in English, but in Dutch you should write the preposition (and the article den) in lower case.

3. word groups: in the source of the Dutch dictionary there are many word groups (borrowed from English, French, Latin ...: new age, déjà vu, in dubio) but just like hyphenated words these aren't regarded as a whole. To tackle the problem, the parts of these word groups are added to the dictionary. But that has got its side effects: in Dutch we write seks for English sex, but in safe sex you should write an x; by including sex in the Dutch dictionary Opera (and others) can't detect that word as a (very common) spelling error. To my knowledge Hunspell doesn't offer anything like "sex (only correct if preceded by safe)". If we include both British Columbia and Colombia in our dictionary, would it be possible to mark British Colombia as an error?

4. dots and periods: Opera but not OpenOffice.org nor Firefox has got problems with a dot/period in the middle of a sentence. Actually OpenOffice.org is in the default dictionary, but Opera doesn't see it as a whole and unbelievably proposes OpenOffice.org as the correction for OpenOffice. Probably Opera considers the dot as the end of the sentence, even if the next word is in lower case.

Hopefully the Opera spell checker can be improved in the following months!

I think there are some strange things going on.

In the case of the Australian English dictionary which I created, here is what I find which corresponds to the points you raise.

1. Hyphenated words are checked as a whole. The test I use is bric-a-brac. Bric-a-brac is in the dictionary and is thus not an error. Neither the terms bric and brac are in the dictionary and show up as errors as they should. This is the opposite to what you are finding and that is weird.

2. The Hunspell dictionary which is based on MySpell takes into account capitals. For example my service is called JustLocal. This is included in the dictionary since I'm the maintainer. As you can see it has a capital J and capital L. This word is handled correctly and is not shown as a spelling error. The words justlocal, justLocal are as expected shown as spelling errors. Again this is not what you are finding which is weird.

3. Now spellchecker I found can handle compound words which contains a space. A space is often used to delineate words. I wish there was a way to handle such words. At best both are entered into the word list (or not entered at all) which is obviously not ideal and leads to potential errors. A solution is required for this problem.

4. Agreed that periods are not handled correctly by Opera 10 spellchecker.

I'm providing this feedback as it shows Opera 10 is working in some languages, but not others.

@JustLocal,

Thanks for your clarifying reply.

I can understand that Opera's spell-checking is better for English, but hope that support for less used languages will improve. After all, the problems mentioned in my previous post aren't unique for Dutch, but seem to apply to many (?) other languages.

The word bric-à-brac (with the French accent) also figures in the Dutch dictionary and is valid, as it should be.
But so is the word bric, which is not a valid word but is in the Dutch dictionary (probably for Hunspell reasons).
Strangely enough, brac happens to be correct too, though it's not in the dictionary! Does Opera extract it from the dictionary entry bric-à-brac?
And finally bric-a-brac also passes, but unlike Australian English needs the accent on the a.

Indeed, the implementation for Dutch (and other similar languages?) is far from ideal.

I think the newest Hunspell version can handle something like REP alot a_lot (see http://sourceforge.net/docman/display_doc.php?docid=29374&group_id=143754 ). Couldn't the underscore be used to include entries as déjà_vu or New_South_Wales to correct mistakes like déjà vue or New South Whales (which according to the default dictionary seems to be correct)?

Jan

I like the Opera.com visual sample of misspellings:


(I hope the moderators don't mind me hotlinking from their site.)

This model, obviously taken from Windows Vista, is very good. For starters, one pixel always separates the base of most letters and the top of the "misspelling wave", as I like to call it. The spelling wave alternates between its top and bottom "sections" by a consistent two horizontal pixels per section. In addition, the wave is a total of two pixels high from the top section to the bottom one; I like how proportional the wave looks along with the text.

Here's a screenshot of my implementation on Windows XP SP2 (32-bit):


Here, the red line is very haphazard in implementation. The spelling wave collides with the base of most words. The wave's sections consist of two horizontal pixels on the top section and three horizontal pixels on the bottom one. In addition, the wave is a total of four pixels high from top to bottom section, a bit too overwhelming for the words themselves.

Ideally, below 100% zoom, the wave shouldn't change in size or implementation compared to when set at 100%. Past 100% page zoom, the misspelling wave should proportionately increase in size with the text, to the nearer pixel.

In addition, the spell checker, by default, lists the following words as misspelled:

QuickTime
Macromedia
shockwave
plugin
plugins
webpage
screenshot
screenshots
frontpage
BitTorrent
userbase
hotlink
hotlinking
toolbar
toolbars
tri-license
StumbleUpon
Wikipedia

@mikemikeb

Though I prefer Opera's bold waves to Firefox's faint dotted lines, the waves are rather ugly.

Indeed, if the top of the wave is under certain letters, it gets difficult to distinguish them: in behavoir (see your screenshot titled Quick reply) the v can easily be confused with a y.

There is another practical proposal which I take from your screenshot titled What are you doing?: the suggestion window should not be at the left nor the right of the misspelled word, and it should never cover that misspelled word.
The suggestion window should be under the spelling error (or maybe above it if the error is at the bottom of the page) so that you can read the complete sentence.
That's the way OpenOffice.org does it.

The wavy line thing has already been discussed at some length. No doubt it will be improved.

Wikipedia, StumbleUpon, QuickTime, Macromedia, Shockwave, Frontpage, BitTorrent are all proper names, so it is debatable whether they should be in any dictionary by default.

plug-in, plug-ins. screen-shot, screen-shots, user-base hot-link, hot-linking, should be hyphenated

webpage should be web page

toolbar, toolbars should be in most dictionaries by now. I have added them.

tri-license should be tri-licence, but its not a word that I know.

Originally posted by Pesala:

webpage should be web page

Even better would be to write Web with a capital letter.

Safari, for a difference from last test, was now very fast to.

Once upon a time their was a little fairy hammy named Kala.....

Like most people I like Opera and I like everything witch is connected with it.

Just three quotes taken from the internet. I suppose many of us (especially if English isn't your mother tongue) make these kinds of spelling errors... and the (Opera) spell checker will be of no help in these cases because to (instead of too), their (instead of there) and witch (instead of which) are all valid words in English.
Does anyone know if there is a spell checker which can deal with these errors?

I know a Dutch spell checker that has got a lot of these 'false friends' in its database. They could be marked in some way inviting you to hover over them and then explaining the difference between witch and which (e.g. as a tooltip).

Yes, doubleclicking on a word lets you look up that word in a dictionary or an encyclopedia. Perhaps that could help from time to time.

Originally posted by Heureka:

Safari, for a difference from last test, was now very fast to.

Once upon a time their was a little fairy hammy named Kala.....

Like most people I like Opera and I like everything witch is connected with it.

Just three quotes taken from the internet. I suppose many of us (especially if English isn't your mother tongue) make these kinds of spelling errors... and the (Opera) spell checker will be of no help in these cases because to (instead of too), their (instead of there) and witch (instead of which) are all valid words in English.
Does anyone know if there is a spell checker which can deal with these errors?

Those are not spelling errors, they're grammar errors, much harder to deal with for computers. Example, if you type haar haar was blond (her hair was blonde) into office it will mark the second haar incorrect as a repetition, while the sentence is of course perfectly correct, the two haar' s have different meanings. I know of no independent spell checker that can deal with grammar properly. There is an extension for Openoffice that tries, but it is useless.

Originally posted by Heureka:

I know a Dutch spell checker that has got a lot of these 'false friends' in its database. They could be marked in some way inviting you to hover over them and then explaining the difference between witch and which (e.g. as a tooltip).

Which is that?

@Zotlan

You are right, some spelling errors can be considered as grammar errors.
As a Dutchman you probably know errors like ik vindt (should be ik vind), hij word (should be hij wordt), ze hebben het verandert (should be ze hebben het veranderd) which most spell checkers can't detect, though the internet is full of them.

But let me give you another example:
Emerson won the price for poetry in 2006.
Here the words price and prize are interchanged, but both are nouns, so the sentence remains grammatically correct. It's a question of semantics, not of grammar (or syntax).

The Dutch spelling checker I mentioned in my previous posting is Van Dale Spellingscorrector, which I believe was developed by TNO. See http://www.hcw.be/p.aspx?p=H1820 and http://www.kennislink.nl/web/show?id=77489 for more information.

It would be great if Opera did grammatical checks also--but software that both checks grammar and spelling is pretty rare (I only know of Microsoft Office). I prefer to remain realistic and thankful that we at least have a spell checker now

Originally posted by Heureka:

The Dutch spelling checker I mentioned in my previous posting is Van Dale Spellingscorrector, which I believe was developed by TNO. See http://www.hcw.be/p.aspx?p=H1820 and http://www.kennislink.nl/web/show?id=77489 for more information.

It sounds impressive (though it is disappointing that it only seems to work with Microsoft products). However, products like this would be expensive for Opera to include, i do not think there is anything like it in the open source world. I think there are better places for Opera to spend its resources.

Originally posted by Pesala:

tri-license should be tri-licence, but its not a word that I know.

Hunspell uses the term tri-license as per their site "C++ library under GPL/LGPL/MPL tri-license." I hope that helps.

Somewhat ominously, the very obvious bold squiggly issue at 100% was not remedied in the latest build, which is 80 builds newer than the original.

Don't expect your favourite bug to be fixed before anything else. There are hundreds of more important things than the style of a squiggly red line. I expect it will be fixed eventually.

It's only my 12th favorite bug, but I think the larger point is, as usual, thanks to Opera's crack bug reporting system we don't even know if they consider it to be a bug in the first place.

Originally posted by rseiler:

It's only my 12th favorite bug, but I think the larger point is, as usual, thanks to Opera's crack bug reporting system we don't even know if they consider it to be a bug in the first place.

Yeah, it's really annoying that you take the time to report an issue, and then you're kept in the dark. I think people would complain a lot less if they were more open.

Originally posted by Pesala:

tri-license should be tri-licence, but its not a word that I know.

That depends on which version of English you use.

Originally posted by BtEO:

That depends on which version of English you use.

You are certainly referring to the differences between British and American English

license^AE, licence^BE [comm.]

-- as shown in LEO-dictionary e.g.

Hello.
I'm french.
French dictionnary works fine but is it possible to control text area without entering in the field ?
The dictionnary works only if I click in the field and it's not convenient for me.

Thanks
Olivier

You can make spell-check work on a non-text area by entering design mode. CheckWebPage.png

Edit Drag the link to a toolbar to create a button. Click to toggle design mode on/off.

Originally posted by sopheos:

French dictionnary works fine but is it possible to control text area without entering in the field ?
The dictionnary works only if I click in the field and it's not convenient for me.

No. This doesn't seem to be possible.

Pesala! Your Edit button is a pearl! It's great to spellcheck self-made pages, too...

The only problem I have with the spell checker is that it... well... sucks at spell checking. Mistakes that have a misplaced letter in the middle, or a missing apostrophe are never recognized properly.

Originally posted by veggen:

The only problem I have with the spell checker is that it... well... sucks at spell checking.

I presume that the suggestions list will be fixed. If not, then the spell-checker is useless. Already noted in this post.

The spellcheck is a gr8 feature, though mine only works with a new email. If I reply or forward an email, whatever I write is not spellchecked .....

I find the spell checking feature quite great. It's already saving my life while writing for my blog. However, I have one request/idea/suggestion.

Since I create multi-language web sites, the current implementation is a bit annoying. I continuously switch from one language to another. However, there should be a very simple way to choose language. Any half-correctly coded web page should use the LANG attribute to identify its language (applicable to HTML, BODY or most divisions of the page). Having the choice of the language is great, having the option to use the language set for the text area or the page auto-selected would be even better.

Auto-selection of the dictionary could work like:
- if text-area (or higher hierarchical division) has a LANG attribute, look for the correct spell-checking dictionary (same name as LANG attribute: en-EN or fr-FR or fr-CA, etc.), if not fall back to root LANG (en or fr, etc.), if not fall back to default language of the browser.


To me, it seems easy and natural. But maybe I'm wrong and over-optimistic.
Anyway, it would be great since I could jump from page to page without needing to switch from one dictionary to the next (I tend to write simultaneously one article in English and its translation in Frnech (or the other way around)).

What do you think about this?
Yves

Originally posted by Roumazeilles:

if text-area (or higher hierarchical division) has a LANG attribute, look for the correct spell-checking dictionary (same name as LANG attribute: en-EN or fr-FR or fr-CA, etc.), if not fall back to root LANG (en or fr, etc.), if not fall back to default language of the browser.

I think it's a clever idea, but perhaps abuse of the standard. The lang attribute is property used by the user agent for display purposes. It's not meant to control the language you enter into a form. I'm not sure people would appreciate Opera telling non-Americans they spelling stuff wrong because they are on an American site and force the user to switch the dictionary back to their language. Most people only speak and spell in one language so auto-switching would accommodate only an extreme minority.

Having said that, the latest build allows you select your language very quickly from the context menu.

Originally posted by drworm:

I think it's a clever idea, but perhaps abuse of the standard. The lang attribute is property used by the user agent for display purposes.

That may be the case (I'm not an HTML specialist). But, quoting from http://www.w3.org/TR/REC-html40/struct/dirlang.html :

Some situations where author-supplied language information may be helpful include:

• ...
  
• Assisting spell checkers and grammar checkers
  

Which would quite warrant my suggestion as compatible with the use of the LANG attribute.

Anyway, apart from this theoretical approach, I get the practical feeling that if the page/text-area of a web site is tagged as German, then most text is expected to be in German, including in the comments boxes or other text-areas. Checking the web with Google would probably prove me right.

However, the point may be a little moot, because I have an idea that would not interfere with any other current usage of the spell-checking feature. As you point out, the latest version gives a very fast way to switch from one language to another from the right-click. It's this part that I would like to see improved, but certainly not in reducing the confort of other users. More precisely, my suggestion would be to:
1. list the dictionaries found by Opera 10
2. offer an additional option: auto-select from page attributes
3. let the user choose

For most people, it would change nothing from the current status of features (an additional option that they may naturally ignore). For people using heavily the multi-tab feature, with web sites of different languages and/or multi-language web sites, it would bring an enormous improvement in terms of fast-switching from one context to the next.

Furthermore, while HTML 4.01 allows to locate the LANG attribute on nearly all the language elements, it could be first tested by detecting the LANG attribute on <HTML> or <BODY>. Only in a second step would it be necessary to track the complete implementation of LANG (theoretically, you can hierarchically go from one LANG to another, like with a Japanese word/sentence isolated in a Spanish web page). I don't know the organization of the Opera software, but tracking exhaustively this attribute may not be fully implemented today. Until then, an alpha version could easily work with a partial implementation aimed at UI validation.

Just a thought I'm trying to elaborate.

Choosing a word from context menu should rise onchange listener to make sure letter-counting scripts show correct message length (checking if message fits in allowed length).

Originally posted by AKA Panama Jack:

Is there a way to turn off the in-line spell checker and turn ON the ASpell spell checker

I also would like this option as I prefer ASpell's dictionary and I do not like in-line spell checks and never have liked in-line checks in any programs.

I wouldn't care to install Aspell if I could turn off inline checking and check spelling with Hunspell via a dialogue box as I can in OpenOffice.

If only there was a way to know whether Opera considers the thick wavy line problem (seen right here in this very forum) a bug or not. Oh, well, we're just the users, we probably shouldn't bother our pretty little heads with such details.

Originally posted by rseiler:

If only there was a way to know whether Opera considers the thick wavy line problem (seen right here in this very forum) a bug or not.

For goodness sake stop whining about the red wavy line. We are only a few weeks into testing.

Originally posted by daroc:

Choosing a word from context menu should rise onchange listener to make sure letter-counting scripts show correct message length (checking if message fits in allowed length).

If you're sure about that, then you should submit a bug report, seems rather important.

Originally posted by Pesala:

Originally posted by rseiler:

If only there was a way to know whether Opera considers the thick wavy line problem (seen right here in this very forum) a bug or not.

For goodness sake stop whining about the red wavy line. We are only a few weeks into testing.

I too would like to know if this is considered to be a bug or not.

Originally posted by Luchio:

I too would like to know if this is considered to be a bug or not.

It has been reported as a bug, hasn't it, so its a bug. You should know better than to expect an answer on specific issues — what its priority is, etc.

Originally posted by Roumazeilles:

Originally posted by drworm:

I think it's a clever idea, but perhaps abuse of the standard. The lang attribute is property used by the user agent for display purposes.

That may be the case (I'm not an HTML specialist). But, quoting from http://www.w3.org/TR/REC-html40/struct/dirlang.html :

Some situations where author-supplied language information may be helpful include:

• ...
  
• Assisting spell checkers and grammar checkers
  

Which would quite warrant my suggestion as compatible with the use of the LANG attribute.

I stand corrected. For the arguements I said, I didn't think it would be suitable. I guess for most sites you use, you can choose your language/region as a registered users, so the lang property on a textarea can be set specifically for you.

Originally posted by Pesala:

It has been reported as a bug, hasn't it, so its a bug. You should know better than to expect an answer on specific issues — what its priority is, etc.

For goodness sake, is that supposed to be a joke? Just becomes something is reported does not make it a bug. Of course.

And, actually, on more than one occasion I've seen someone with access to the system swoop in here and impart the status of something.

The alternative is to shut up about it, as you suggest, stop "whining" about it (which, I guess, means mentioning it twice in the six weeks since testing started) and then eventually get to the point in testing where only "critical" bugs get fixed, at which point you're forced to wait until next time.

Originally posted by drworm:

For the arguements I said, I didn't think it would be suitable. I guess for most sites you use, you can choose your language/region as a registered users, so the lang property on a textarea can be set specifically for you.

I guess that this is exactly my point. The LANG property is actually either set on a per-textarea basis or a per-page basis (more commonly on per-page, I think; But I may be wrong, I have no stats). Since the LANG property gives the information about which language is used at a specific location, it should be feasible and useful to be able to use this information to select the dictionary to be used by Opera 10.

So, as you said, I can choose the language (as a registered users, or not) and the LANG property is set. It's just that Opera 10 (and neither FireFox) does not follow in this direction and does not use this information provided freely that could support a better spell-checking.

We'll see if this idea catches the attention of some opera developer...

Originally posted by Roumazeilles:

We'll see if this idea catches the attention of some opera developer...

You might want to post a new topic in the Wish-list forum for the idea.

Originally posted by Pesala:

Originally posted by Luchio:

I too would like to know if this is considered to be a bug or not.

It has been reported as a bug, hasn't it, so its a bug. You should know better than to expect an answer on specific issues — what its priority is, etc.

I'm sorry if you feel that your thread is being polluted by this issue, it wasn't intended.

I don't /expect/ an answer and certainly don't demand one, I just /hope/ for one. They might consider thick wavy lines to be perfectly acceptable behavior. We don't know.

Originally posted by Luchio:

I'm sorry if you feel that your thread is being polluted by this issue, it wasn't intended.

The whole Beta forum is polluted by this topic. People keep asking for feedback on bug reports, but the developers do not respond unless they wish to. If you had seriously intended to find out about feedback from the developers you would have posted a new thread on that topic or posted in the blogs, but like everyone else who asks for more feedback, you're just whining. If you had some experience of beta testing you would understand why its no use asking for feedback. It is called a non-disclosure agreement.

Of course this is a bug. The red line suddenly gets much thicker at certain zoom levels.

hi, I upgraded from 9.62 to 10 alpha. Spellchecker doesn't work. I enabled "Spellcheck enabled by default" and tested with other languages. I think it's because of upgrading; Anyone has a solution?

@P30Carl,

Though Opera 10 seems to be fairly stable, you should not install it over Opera 9.62 or 9.63, but in its own folder (by default: Opera 10 Preview). Did you try that?

Spell-checking in English (and in Dutch, my mother tongue) works well then.
Some OpenOffice.org dictionaries (e.g. Hungarian) don't seem to be compatible with Opera 10, but Farsi seems to work: I downloaded the Farsi dictionary and unzipped it in the appropriate folder, then I copied the Teheran article from Wikipedia in Farsi, changed the spell-checking language to Farsi (which I don't understand) and most of the words appeared to be OK.

Good luck!

Originally posted by Pesala:

The whole Beta forum is polluted by this topic.

If it is, do you have any doubt why that is? With a broken feedback system, that's what happens. Opera could have solved the problem years ago but chose not to. Something about making your bed and lying in it comes to mind.

People keep asking for feedback on bug reports, but the developers do not respond unless they wish to.

Profound. How about this: If no one asks, there's even less of a chance of an answer.

If you had seriously intended to find out about feedback from the developers you would have posted a new thread on that topic

Let me make sure I understand you. It would be better to open entirely new and redundant threads (which somehow isn't polluting), referring to current threads, than asking for clarification on an issue already discussed in an existing thread? That sounds like a plan.

but like everyone else who asks for more feedback, you're just whining.

Uh huh. What a charitable definition of the word you have.

If you had some experience of beta testing you would understand why its no use asking for feedback. It is called a non-disclosure agreement.

Yikes. You actually think people in the know don't pop in here very often because of NDAs? That's incredibly unlikely, if for no other reason than it doesn't explain when people DO.

Of course this is a bug. The red line suddenly gets much thicker at certain zoom levels.

Um, but which one is correct? Do you know for sure? Some sites it's thick, some sites it's thin -- at the same zoom level. Some people think thick is perfectly fine, others not. I ask again: how do you know?

Originally posted by rseiler:

If it is, do you have any doubt why that is?

It is because ignorant people like yourself have no respect and do not understand how beta testing works. The problem lies at your end — you're just too thick to understand how to report bugs and then just get on with your life, while the Opera developers get on with what they're paid to do — fixing bugs instead of responding to ignorant users in forums to explain what is a bug, and what is by design; what is high priority and what is low priority. Yes, I am sure that the red wavy red line is not supposed to double in size between 200% and 220% — therefore it is a bug. Exactly how thick the line should be is obviously down to personal opinion. You cannot please everyone.

Originally posted by rseiler:

It would be better to open entirely new and redundant threads (which somehow isn't polluting), referring to current threads, than asking for clarification on an issue already discussed in an existing thread?

This thread is about the inline spell-check. Discussions about the bug reporting system do not belong here. That is a general topic which belongs in the blogs or the feedback forum, and it has nothing to do with any particular bug. All off-topic discussion about the bug reporting system is just polluting the forum, and making it harder for the developers to focus on the issues that obviously do still exist in this brand new, and therefore unpolished Alpha version of Opera.

Originally posted by Heureka:

you should not install it over Opera 9.62 or 9.63

I didn't know this before installing alpha. and when I knew that, I tried to uninstall that; but it didn't uninstall. I have tried many ways and the problem is still unsolved.

I know how to use spell-checker, but I can't do that.

It's great to see inline spell-checking coming to Opera. May I suggest (beg?) that you support the DOM-level disabling of this feature like Gecko does? In Gecko I can prevent the inline spellchecker from examining the text nodes in any DOM subtree by setting a parent block element to spellcheck = 0.

This is not as esoteric a request as it may seem -- some web pages want to do their own spell-checking for whatever reason, and this is the only way to prevent double-underlining and double-suggesting from occurring. Also, the WYSIWYG editors sometimes expose this feature: e.g., TinyMCE offers a gecko_spellcheck option in its config. If Opera 10 were to support the same mechanism as Gecko (node.spellcheck = 0) then the existing code to disable Gecko inline spellcheck would automagically work for Opera10, which would be sweet.

Thanks for listening.

I read all 147 posts here and i still don't know if there's a way to disable the inline spellchecker (most of the times it's useless) and use the Aspell instead (which works perfectly in my default 9.63). Is there some tweak that would make this possible ?

Originally posted by Azrael1:

I read all 147 posts here and...

No, you missed post #1!
Spellcheckenabledbydefault
Disable this...
... a somewhat silly designation however, IMO... :-(

BTW, don't forget to SAVE!

Originally posted by HaJotKE:

No, you missed post #1!

And you misread Azrael1's post apparently. He/she wants to use Aspell instead of Hunspell.

I hope that dialogue checking will be enabled for Hunspell as it is in OpenOffice. Using a dialogue is much more efficient.