[Interest] Input method support in Qt 5

Dominik Wezel dom at qvasartech.com
Tue Nov 8 20:28:33 CET 2016


Hi André

I must apologize: you are right, and /home/user/.XCompose actually 
works.  One thing I didn't notice until now was that together with the 
upgrade to KDE 5, all Compose files on my local machine were overwritten 
with their original versions — a very rude thing to do to a user who 
already suffers from regular overrides of his stunnel and sshd conf … 
but I won't complain, because the upgrade was an operation bordering on 
giving birth to sextuplets, and maybe I made some mistakes myself.  So 
what I actually copied to /home/user/,XCompose was the standard X11 key 
composition table.  Unfortunately, I only do a very simple 2-generation 
backup, so the original is long lost since then, but thanks God Emacs 
made its own backup copies to Compose~, hence I lost only my most recent 
change.

I'm very glad that this problem and riddle is solved now, and hopefully 
other people can read the solution in the discussion on the Qt interest 
group this thread is hopefully posted to.

Thanks to all

Dominik


> On Mon, Oct 17, 2016 at 12:45:32PM +0200, Dominik Wezel wrote:
>> Hi
>>
>> I had to upgrade to KDE 5 recently.  One of the first things I immediately
>> noticed was that my key composition table was no longer working.  In earlier
>> versions, I managed to force Qt and GTK to use XIM, which supported my huge
>> additions to /usr/share/X11/locale/en_US.UTF-8/Compose.
>>
>> As a mere KDE user without deeper insight, I first blamed KDE for it, but
>> was kindly redirected to Qt as the underlying problem. When searching the
>> net I found some emails from 2013 written by Paeglis Gatis, where I learned
>> that XIM is deprecated in Qt5 and that support for key composition tables is
>> planned to be done directly in Qt.
>>
>> However, it seemed that task https://codereview.qt-project.org/#/c/48185/ is
>> abandoned.
> Note the abandoned reason. This was finally merged as
> https://codereview.qt-project.org/#/c/51555/
>
> Andre'





More information about the Interest mailing list