>
> Because then every time an additional language is added, -every- printk
> must be hunted down and modified.
>
> I think the best solution would be to do something like so:
>
> Have one file, lang_default.h, that liiks like so:
>
No, not one file, bad idea; if that one file is changed, every file that
includes it will be rebuilt (just like the current config.h situation).
> #define PRINTK_PANIC "Kernel Panic"
> #define PRINTK_BAD_IRQ "Invalid IRQ"
> #define PRINTK_HOMICIDAL_CPU "I'm sorry Dave, but I can't do that."
>
> and a series of files, like lang_piglatin.h, that looks like so:
>
> #include <lang_default.h>
> #undef PRINTK_PANIC
> #define PRINTK_PANIC "Ernalkay Anicpay"
> #undef PRINTK_HOMICIDAL_CPU "I'mway orrysay Aveday, utbay Iway \
> an'tcay oday atthay."
>
> kernel modules would use the symbolic versions of the strings
> instead of the original strings, and all is well.
Oh, so we slow modules down, when we could use the same zero runtime
overhead, hmm ...
> A program can strip
> out the strings from the kernel sources to initially create
> lang_default.h, as well as report the locations in the source where the
> strings wrere found (to facilitate conversion).
>
Well at the minimum I'd want one per directry; better one per driver, but
that's non-obvious from a utility searching through.
> Then we can get onto more -important- issues of kernel message
> internationalization, like if color is spelled with a 'u' or not.
>
With ;)
>
> --
> Buddha Buck bmbuck@acsu.buffalo.edu
Bryn
-- PGP key pass phrase forgotten, \ Overload -- core meltdown sequence again :( and I don't care ;) | initiated. / This space is intentionally left | blank, apart from this text ;-) \____________________________________