Re: Kernel-Messages translation (fwd)

Robert Glamm (
Tue, 17 Jun 1997 11:03:31 -0500 (CDT)

> > > One thing: does anyone realize what this is going to mean for bug
> > > reports? I can guarantee that if I receive log messages that I can't
> > > read, I'll hit the delete key so quickly you wouldn't even notice.
> > >
> > > I think this is a bad idea. We're talking about the kernel,
> > > ferchrissake...
> >
> > why not like Oracle: 'ORA-18560: kak tebja zawut'
> > 'LIN-12341: fatalitsisky errorno inodo'
> >
> > also, bug reports would be more consistent, the string 'LIN-' would likely
> > be present in a bug report.
> >
> > [English could be a fallback language for untranslated messages.]
> If (BIG if here) Linus was to ever accept all possable translations for
> each kernel message into every possable language, this is the only sensable
> way to do it. If each message has a unique code (that is independent of
> language) it might be possable. Of course, not all alphabets include
> the required symbols for Latin kernel string identifiers. It would be
> a pain to get bug reports that had the kernel error identifier string in
> some multi-byte hieroglyphics that did not translate to a Latin alphabet
> at all...

Write a small perl script to include in the scripts/ directory that
reads identifier/English pairs into an associative array, scans for
error messages via identifier & replaces them with the english value from
the associative array.

For that matter, I wonder if it could be done automatically as a back-end
processing agent for majordomo on without killing the
load too much... Seems to me that since majordomo is already written in
perl adding a few lines of code to scan & replace ID/error pairs in the
body wouldn't be too much.

"If HP was only an $8 billion  | Bob Glamm  H: +1 612 6239437 W: +1 612 6268981 
 company like Sun, we also     | URL:
 might be less ambitious."     +-----------------------------------------------
 -- HP's Lewis Platt referring to Sun's refusal to support Windows NT