On 2012-09-11, at 7:59 AM, norm(_at_)dad(_dot_)org wrote:
I guess I kinda believe the opposite. For example send won't send a message
if it
doesn't understand any of the addressees. I think that's a good thing. The
general philosophy of mh was (contrary to the UNIX philosophy) that if
anything
is wrong do nothing.
Not knowing a file's MIME type is not an error. It can't be, since not every
type of data has a specific MIME type.
E.g., what is foo.dot? foo.xyz? cat.8?
These are all files that exist on my systems. foo.dot and foo.xyz are text
files. The suffixes are local differentiators with meaning only to me. Adding
the dot and xyz suffixes to mimetypes with a text/plain attribute is wrong,
since another .xyz file might contain, say, sample raw data from a random
number generator. I.e. there is no recognized conventional type for .xyz
files. This is why MIME has fallback rules: text/plain for printable ASCII
text content, and application/octet-stream for everything else.
What about cat.8? On my UNIX systems, that's a pre-formatted man page
(text/plain). On my Plan 9 systems, it's object code from the 80386 compilers.
(Ditto for cat.2, which could be a pre-formatted man page for a system call,
or object code from the 68020 compilers.) This is why deriving the MIME type
from the file name is fatally flawed.
It's not MIME's job to fix your typing mistakes.
--lyndon
_______________________________________________
Nmh-workers mailing list
Nmh-workers(_at_)nongnu(_dot_)org
https://lists.nongnu.org/mailman/listinfo/nmh-workers