Re: uuencode: multi-bytes char in remote file name contains bytes >0x80

[Bruce Korb]

On 07/03/11 04:14, 張叁 wrote:
> my code is just showing my meaning.
> may not works well.

Hi Duhuanpeng,

RE: enhancement to have uuencode encode output file name:

A few other things that will be needed:

1.  changes to mark the file name as an encoded file name
2.  parallel changes to uudecode that will then convert the
    hex encoding of the file name to the file name to actually use
3.  This should be an option to uuencode, rather than a compile time setting.
    uudecode should detect it.
4.  research into uuencode/uudecode to ensure the format enhancement
    is compatible with the POSIX spec (does nothing to violate it):
    http://pubs.opengroup.org/onlinepubs/009604599/utilities/uuencode.html
    http://pubs.opengroup.org/onlinepubs/9699919799/utilities/uudecode.html

My reading of it seems to indicate that there isn't any wiggle room on the
"begin" line, for either base 64 or traditional encoding.  I think it would
require a preamble line that put uudecode into a non-POSIX state wherein
uudecode would know that the input file name was encoded.

Given that the encoded file may get handed off to a uudecode that
knows nothing about this magical state, the file name encoding should
encode '/' characters with the '/' character.  The technique in your
patch simply hex encodes the entire file name string.

I've CC-ed the gnulib list because there are folks there much more i18n
literate there than I am.

Sorry to make it so difficult....

Regards, Bruce