Caret comes to top
Saito
FedoraCore5.
For the Japanese input, the default method of FedoraCore5 is SCIM and
scim-anthy.
On the above builds of Opera, after hitting some keys, use SPACE BAR to
convert
the letters into Japanese characters, hit RETURN key to confirm it, the
caret,
i.e. blinking cursol mark, comes to the top of the confirmed words, not to
the end
of it.
The above issue is confirmed by friends, who use recent Debian, and if you
change
~/.qt/qtrc 's [General] XIMInputStyle=On The Spot into Over The Spot,
then you can't use inline input, but can ecape this irritating issue.
I forgot if I reported this issue before, but at least this week's build
still
has this, so post it here.
--
Saito, a paid Opera user both on Windows and Linux
http://d.hatena.ne.jp/saiton/ - Diary in Japanese
http://my.opera/saito/
Eirik Byrkjeflot Anonsen
> TP2 of Opera9 and recent weekly builds, I use static-qt of Opera9 on
> FedoraCore5.
> For the Japanese input, the default method of FedoraCore5 is SCIM and
> scim-anthy.
>
> On the above builds of Opera, after hitting some keys, use SPACE BAR
> to convert
> the letters into Japanese characters, hit RETURN key to confirm it,
> the caret,
> i.e. blinking cursol mark, comes to the top of the confirmed words,
> not to the end
> of it.
>
> The above issue is confirmed by friends, who use recent Debian, and if
> you change
> ~/.qt/qtrc 's [General] XIMInputStyle=On The Spot into Over The Spot,
> then you can't use inline input, but can ecape this irritating issue.
>
> I forgot if I reported this issue before, but at least this week's
> build still
> has this, so post it here.
>
We think input methods and dead keys will work again in the shared
version now! We have finally fixed the problems with qt-immodule. I
believe the fixes are in build 267 and later.
While fixing this we also discovered the reason for the bug mentioned
above (caret is left at the beginning of the inserted text), and we
have fixed that as well.
eirik
Saito
Thanks Eirik and Opera Linux team, for fixing long awaited issue.
This mid weekly, build 272, shared-qt, can accept Japanese input.
But,... it sometimes doesn't accept any, yes any key input at all.
I can just input 'g' at URL bar, and nothing any more :(
It needs to reboot Opera for several times, to get working one.
Eirik Byrkjeflot Anonsen
> On Thu, 11 May 2006 16:56:53 +0900, Eirik Byrkjeflot Anonsen
> <ei...@opera.com> wrote:
>
>> We think input methods and dead keys will work again in the shared
>> version now! We have finally fixed the problems with qt-immodule. I
>> believe the fixes are in build 267 and later.
>>
>> While fixing this we also discovered the reason for the bug mentioned
>> above (caret is left at the beginning of the inserted text), and we
>> have fixed that as well.
>
> Thanks Eirik and Opera Linux team, for fixing long awaited issue.
> This mid weekly, build 272, shared-qt, can accept Japanese input.
> But,... it sometimes doesn't accept any, yes any key input at all.
> I can just input 'g' at URL bar, and nothing any more :(
>
Do you have any idea when this happens? Does it (only) happen when
the 'autocompletion' box shows up? Does it (only) happen while typing
in the address field on top of the web page? How about dialog boxes?
eirik
Saito
<ei...@opera.com> wrote:
Now I can use my keyboard. This is my second try :)
If "it" happens, any input is impossible, ie, I can't
write mail or bloggings. Once it happens, I have to restart
Opera, then check if next started Opera is working or not.
And once it works, during that session, it sometimes happens.
So in that case, I have to restart Opera itself.
I don't see any reason. One of my friends, using Fedora Core 5,
abandoned share-qt, and chnaged back to static-qt again.
Eirik Byrkjeflot Anonsen
It would be very helpful if you can find out something about what you
do just before it happens. Do you think it only happens while writing
japanese? Or does it happen even while you're not typing anything at
all?
eirik
Saito
<ei...@opera.com> wrote:
OK, I seem to detect the pattern of it.
While I switch IME on, I often type Japanese in URL bar.
Then I delete the words typed there and switch IME off,
to try to type "g opera desktop" or something, "it" happens.
No more key input at all after that.
Saito
> OK, I seem to detect the pattern of it.
> While I switch IME on, I often type Japanese in URL bar.
> Then I delete the words typed there and switch IME off,
> to try to type "g opera desktop" or something, "it" happens.
> No more key input at all after that.
Just a shot addition to my previous post.
I don't have to switch on and off IME, if I just
open,foucus on URL bar and try to input something there,
"it" happens.
Separate Google Search field works. To try to "g somethig" is NG.
Saito
> On Fri, 12 May 2006 17:54:37 +0900, Saito <sa...@operamail.com> wrote:
>
>> OK, I seem to detect the pattern of it.
>> While I switch IME on, I often type Japanese in URL bar.
>> Then I delete the words typed there and switch IME off,
>> to try to type "g opera desktop" or something, "it" happens.
>> No more key input at all after that.
>
> Just a shot addition to my previous post.
> I don't have to switch on and off IME, if I just
> open,foucus on URL bar and try to input something there,
> "it" happens.
>
> Separate Google Search field works. To try to "g somethig" is NG.
Another additions.
When you type 'kde' first, then back to top and add 'g' in front of
'kde', ie 'g kde' it keeps working. But if you type 'g' first, then
no key entry is accepted afterwards.
Eirik Byrkjeflot Anonsen
> On Fri, 12 May 2006 18:09:08 +0900, Saito <sa...@operamail.com> wrote:
>
>> On Fri, 12 May 2006 17:54:37 +0900, Saito <sa...@operamail.com> wrote:
>>
>>> OK, I seem to detect the pattern of it.
>>> While I switch IME on, I often type Japanese in URL bar.
>>> Then I delete the words typed there and switch IME off,
>>> to try to type "g opera desktop" or something, "it" happens.
>>> No more key input at all after that.
>>
>> Just a shot addition to my previous post.
>> I don't have to switch on and off IME, if I just
>> open,foucus on URL bar and try to input something there,
>> "it" happens.
>>
>> Separate Google Search field works. To try to "g somethig" is NG.
>
> Another additions.
> When you type 'kde' first, then back to top and add 'g' in front of
> 'kde', ie 'g kde' it keeps working. But if you type 'g' first, then
> no key entry is accepted afterwards.
>
I just thought I'd mention that I think I know what the problem is.
It seems that scim and xfree/xorg does not read the XIM specification
quite the same way.
I think the problem is caused by changing input context while X is
handling a message from the XIM server (i.e. scim). Scim will not
handle any more messages from the same client until X has sent a
SYNC_REPLY back, but X isn't going to do that until X tries to handle
another message for the same input context.
I'm not completely sure this is what happens as I'm no expert on the
internal xorg code, but that's what it looks like to me. I even
managed to reproduce the problem in firefox using this theory... I
don't think there are any simple work-arounds we can do in opera, so
this bug is likely to stay with us for a while yet.
eirik