Hi, On Sun, Jul 16, 2023 at 07:32:03PM +0200, Dragan Simic wrote: > I'd like to implement support for an additional, alternative way for the > color configuration in the dmesg utility. It would use an environment > variable to store the color configuration, in a way rather similar to what > grep(1) does using the GREP_COLORS environment variable. frankly, terminal-colors.d has been implemented to avoid environment variables :-) Anyway, I have nothing against support for <util>_COLORS= if it will be implemented into lib/colors.c and the code will work together. For example if terminal-colors.d/*.disable will force application to ignore <util>_COLORS= too. The ideal solution will be if the env.variable will be just another way how to specify color scheme (alternative to .scheme file). Maybe also follow conventions used by terminal-colors.d, it means scheme, enable and disable options, for example: DMESG_COLORS=disable DMESG_COLORS=enable DMESG_COLORS=alart:37;41,warning:37;42 (not sure about the format, this just brain-storming ... ;-) Karel -- Karel Zak <kzak@xxxxxxxxxx> http://karelzak.blogspot.com