Comment 230 for bug 1245473

Revision history for this message
In , kolya (mar-kolya) wrote :

Dear xorg developers. This is an old bug. In fact this is a 13 years old bug. And it has been at least 7 years since this bug had a patch to fix it.
Yes, existing patch breaks specification. But one would seem that the fact that patch existed for 7 years and was applied by default by popular distributions without users` complains would suggest that existing patch is a practical solution to the problem despite breaking theoretical (and as discussed here - not well thought through) specification.
There is now once more spike of activity here - probably because Ubuntu 17.04 was released recently which dropped this patch because it was not compatible with newer xorg. kyak has provided an updated patch, but it would take a lot of time for distributions to pick it up and then circle would repeat with next xorg version.

Yes, there has been a few heated discussions around this patch and XKB specification. And yes, specifications are important, but this is really annoying problem for people with more than one keyboard layout. Annoying to the point of making xorg really unusable for some groups of people. So is there any chance to get more practical with this bug?

For example, kyak's patch has a function 'xkbSwitchGroupOnRelease' that is a stub to make new behaviour controllable by configuration. Would xorg maintainers find this patch acceptable to upstream if that function read value from some environment variable and only turned new behaviour on if it was set?

I feel like it would allow XKB spec to stay and also would allow users affected by this problem to solve it without recompiling xorg - which average user would struggle to do.

Looking forward to your reply.
Thanks!