On Mon, 2006-04-24 at 15:19 +0900, Tokio Kikuchi wrote:
Well, it should be a mess. :-(
I'm hoping we can make it less so!
Consider mailman get a spam from a foreign country and caused an error. Mailman may complain UnicodeDecodeError and spew an excerpt containing unknown charset string. This is certainly not printable if there is no encoding which means only us-ascii is accepted for the log file. Even if you determine the charset for your language (eg. euc-jp for japanese), you still get error for a chinese spam.
It may be useful if the log output use 'replace' feature of encode() method.
That's probably a good idea. Also, I'm wondering if we should allow users to set the log file encoding in Defaults.py, or whether we should force utf-8, or try to interrogate the system for the encoding value.
Basically, the logger should be as liberal as possible, just in case we let encoding problems slip through (more on that in the next follow up).
-Barry