patch tracking issue

[Mauro Carvalho Chehab]

Em 15-06-2011 10:11, Jeremy Kerr escreveu:
> Hi Albert,
> 
>> (and I'll post an UTF-8-to-ISO-8859 fixup patch at the next merge window)
> 
> Now, that's where it gets tricky - the patch you send will have *both*
> encodings in it.
> 
> I'd suggest checking patchwork after sending it. Or even sending as a
> pull-request.

Jeremy,

The proper fix is that patchwork should not try to detect or change the charset
encoding. It should simply keep the patch as-is.

We're suffering for a long time with all those charset chaos for a long time,
and no matter how many times someone tried to fix, there are always some situations
where it fails by either rejecting an email or by generating messages with 
wrong charsets. I have several complaints from people with accents on their names
that patchwork does the wrong thing and I need to manually fix it, or the
patch enters into the Kernel with the wrong name.

The solution seems to be convert all strings into python to something else where
Python won't try to interpret a charset inside it (or to port it into a smarter
language like C or perl that is not dead-brained with respect to strings).

Just my 2 cents.

Mauro