Messages in this thread | | | Date | Sat, 18 Jan 1997 21:10:16 -0800 | From | Steve VanDevender <> | Subject | Re: Jive -> Kernel (International Linux) |
| |
Daniel G. Linder writes: > I think that Mr. Ross' usage of the individual files works well. The > only thing we will have to get use to is doing printk(HELLO) as opposed > to doing printk("literal string"). How about printk's that have values > in the string? I.E. "CRC sum on packet %i should be %i but is actually > %i\n". In retrospect, these #defined constants are transparent to the > compiler so this should still work, right?
Except that in other languages the word order of sentences is different, so the format items may not be used in the same order in a well-formed sentence, or require a very awkward phrasing in the other language to keep the same order as they would be used in English.
As someone who's done software internationalization before, I can assure you that the problem is not nearly so easy to solve as is being supposed, especially since internally the kernel was never designed for internationalization from the beginning. Having done several such retrofits, I can assure you they are utter hell. I can send interested parties a document I wrote for a former employer detailing the conversion problems I had with a program substantially less complicated than the Linux kernel.
Probably the best argument against internationalizing kernel messages is that currently they are standardized and understandable by all kernel developers. While English is certainly not the native language of many Linux users, imagine the difficulty Linux developers would have with having to know and understand all of the languages used for Linux kernel messages submitted in bug reports. It would also vastly complicate system administration and monitoring software that needs to parse kernel messages and would have to deal with all the different languages.
| |