I wrote remeber in a post.
did a spell check. After i corrected remeber to remember it wrote remeremembernd
wtf is bernd? It seems not only the correct and the false word being mixed. it added a "n" to the word.
... tired it again.
now it was:
remerememberes
remerememberit
remerememberlong
remeber -control
remember -experiment
EDIT:
worked fine... YOU ARE A LYING...
worked fine...
Now it does for me too. Have too check at work again.
Maybe it happens only on my work desk.
YOU ARE A LYING...
I hope you don't mean that serious.
Just becabecauses not happening to you, does not mean I'm liar.
EDIT
THER it happened again!
it changes becauseit's to becabecauses.
On very rare occassion, I've noticed the same behavior. The spell check is a big hack and relies on indexes that are somewhat hard to determine due to the way it tries to ignore things that don't need to be spell checked (quotes, code, urls, etc). But I won't look into it unless you can supply a sample text that always triggers the problem on a given browser / operating system combination.
But I won't look into it unless you can supply a sample text that always triggers the problem on a given browser / operating system combination.
Yea. I understand that.
I tested it a few times with remeber. And because it always did the same i thought it was a reproducable bug.
But now i know it isn't.
test 
remeber
becauseit's
[EDIT]
ok. it just did nothing
Pfft, you should be using this.
Marcello
That's cool, but the amount of requests it generates is ridiculous. If someone wants realtime spellchecking they should use a browser that has built-in spell check.
they should use a browser that has built-in spell check.
Like konqueror.
Or Firefox 2.0.
How many requests does it do? You could wait for a few seconds after they stop typing, then just send the new words for checking. It could be done efficiently, I imagine. That version runs pretty cleanly for me.
Marcello
It triggered thirty-five requests when I typed:
This is a demo of the Lite Spell Checker (plain text implementation).
The spelling is checked as you type and misspelled words are highlighted by a wavy red underline. If something looks weird you might have found a bug, remember it's still a beta, in that case click the 'Update/Rescan' button to restore things.
It just seems like too many unnecessary hits for a site with a lot of people typing messages.
Couldn't you use the same structure but just run the spellcheck if one clicks a button?
Couldn't you use the same structure but just run the spellcheck if one clicks a button?
Of course. But there would be no difference between that and what I'm doing now. Whether you wrap <span> or <select> around the word, you still have to deal with the same issues. The <select> approach is just a bit simpler to implement because there is less "magic" going on.
I think the spellcheck bug may have been worsened with the new way whitespace is trimmed for large blocks of text. I'd like to rewrite the thing, but for now I'll just try to patch this bug.