Re: Internationalizing Linux

david parsons (o.r.c@p.e.l.l.p.o.r.t.l.a.n.d.o.r.u.s)
7 Dec 1998 00:44:44 -0800

In article <>,
Horst von Brand <> wrote:

>What do you do with messages that include variable parts? %%BROKEN_DISK
>that translates to "Disk %d is acting up, replace %s as soon as possible"
>is rather useless...

How about:

whinek(IDE(42), "disk:%d,it:%s", <disk>, <replace>);

where the message for english would be:

IDE[42] => "Disk %(disk) is acting up, replace %(it) as soon as possible";

look at if you want to see
a module that I tweaked up to use this message format. I've left the
implementation as an exercise for the reader, but aside from a
brain-popping conversion session to tweak all the messages over to
that format and figuring out how to do the lkm or daemon to hold all
the messages, it doesn't seem like it's as much difficult as
incredibly tedious.

>Please, this is discussed to death each year, the conclusion each time is
>that the enormous effort isn't going to help any, it will just hinder
>development for nothing. Note that all this "standard message identifier"
>stuff _has_ been tried, over and over (IBM, DEC's VMS, ...) and it hasn't
>worked. Not even for English alone.

As an english speaker those codes haven't been tremendously useful,
but that's not horribly surprising. The code HAVE been useful on
some of the times I've been faced with non-english error messages,
but that only after I'd been trained enough to realize that

"%PLONK-0067% Ce n'est pas bizarre. Visitez le volcan, si vous plait."

was actually inviting me to look it up in the wall of manuals instead
of just laughing hysterically at me.

>And nobody will bother (or dare) translate them, so I get to see messages
>in English (written by Alan Cox), in Swedish (by Linus), some in Italian
>(by Andrea), or Bill Hawes in German, and Alexey in Russian... boy, will
>_that_ be fun.

I'd suspect that there would be a lot of pressure to submit at least
the english-language messages in with the code, because a large part
of the developer base has that language as a trade tongue, and doing
translations for other languages will either be motivated by
dedicated volunteers and/or commercial interests. And if there's no
translation and your system doesn't have one of the message modules
loaded, you'll still get some message, even if it's an illuminating

"IDE42: disk:0,it:/dev/hda"

which you can then collect and trade for prizes in the kernel
mailing list or on c.o.l.&tc.

>Worse. In literature a lost cadenza is lost beauty, in technical literature
>a lost cadenza is lost content, maybe critical content. Besides, just as in
>literature a great writer is required to do a first rate translation, a
>first rate hacker will be required to translate technical content
>faithfully. I for one would prefer her to work on the kernel.

If someone wants to work on translations of the kernel messages, or
is paid to do those translations, it seems somewhat counterproductive
to demand that they work on the kernel instead -- some of us don't want
to work on the kernel, but would rather work on other things, and if
we're thwarted on doing those may lose interest and go somewhere else.

david parsons \bi/ I danced for joy when someone else did devfs and saved
\/ the baseline kernel from my, um, gem of programming
skill. Of course it means I'll probably end up
upgrading to 2.2, but sometimes you've got to break
eggs if you want to make bread pudding.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at