I just noticed that emoticon to image mapping in Gajim is done without taking letter case into account. I wouldn't have noticed it, before it caused me confusion during conversation. Trying to remove the lowercase version from showing as an image, I found that there is only the uppercase version in the emoticons.py file.

I tracked down the problem to the line 2558 in /src/gajim.py. It's re.IGNORECASE + re.UNICODE) while it should be re.UNICODE)

I think that compiling the regex with upper/lowercase differentiation gives more flexibility. It would open new possibilities without becoming less intuitive at the cost of adding more strings to emoticons.py.

If there is any interest in changing this, I can adjust the iconset files and prepare a proper patch against current Gajim version.