Messages in this thread | | | From | "Bruce Korb" <> | Subject | Re: Internationalizing Linux | Date | Mon, 7 Dec 1998 11:00:48 -0800 |
| |
> From: Alan Cox <alan@lxorguk.ukuu.org.uk>
>> One more killer issue please: extraction of the strings from the kernel >> almost entirely (external drivers and panic messages excepted). >> Several benefits ON TOP OF translatability: > >Thats part of the same thing > >> 1. Reduced kernel image size >> 2. Reduced space needs for log buffer > >Those are real issues but #3 is stretching it > >> 3. Reduced latency for interrupt level events (printk's).
OK. It is not the end of the "stretch" issues, tho, either :) 4. Events become easily tunable at compile time 5. Events become easily tunable at emit time (in the driver code) 6. Events become easily tunable at display time (the console display program can use options or an rc file or some such to select priorities by category...etc.etc.)
>> get the assistence of several volunteers to clean up the few hundred >> printk's that do not convert automatically. (I already have >> a program written that will convert 15,800 of the printk's, >> but it skips several hundred. That is still well over 95% of them.) > >Thats a good start. I'm not sure I like the output format but obviously >thats an easy bit to change.
Yes, it is relatively easy to change the comment format around. I am only hung up about one issue with it: it needs to not be specific to the event/printk application. I have a "general purpose" tool that extracts things that look something like "/*=type[.group] name\n...\n=*/" and constructs a definitions file for my template instantiation program (autogen). See:
ftp://egcs.cygnus.com/pub/egcs/infrastructure/autogen.tgz
For this reason, I wish to keep the "event" part of the "/*event.group" construct. It identifies it as applying to events, as opposed to some other "type".
>>From my point of view "shortened" (ie your stuff) is a variant on "english" >(we have now). For debugging kernels the full english would definitely be >useful. Plenty of non fatal messages are the not always non fatal..
No argument from me :)
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |