[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Mime types for medical data formats

Am Montag, den 31.03.2008, 15:56 +0200 schrieb Andreas Tille:
> On Mon, 31 Mar 2008, Daniel Leidert wrote:
> > As long as there is no agreement with upstream, you can use the
> >
> > a) x- prefix approach
> > b) vnd. vendor approach
> >
> > Both are described in the MIME specification (IIRC i mentioned the RFC
> > in the discussion). I would suggest the latter for the application/* top
> > type as long as there is no agreement with upstream.
> Well, I think there will be file types where there is not even any "upstream".
> I just thought about whether we as the Debian-Med project could/should
> add some MIME types, test them here and once we are happy about these
> try to propagated them to a place (whereever this might be) to get
> accepted it outside Debian.

You would need to send a RFC to the IETF. There are some articles, how
this works:


> Moreover I wonder whether the file command is able to read the mime-magic
> stuff.

file has its own database. Take a look at /usr/share/file. The file
magic contains a general database and the magic.mime file contains the
MIME database (use the -i switch of the file command). *But* in the
announcement of version 4.24 of file I read, that the databases have
been merged (my understanding of the announcement).

If you want to extend the database, you can create the files /etc/magic
and /etc/magic.mime. The magic(5) manpage contains some useful
information, how the database is constructed. *But* the file database
cannot be extened automatically! There is AFAIK not command to add or
remove some definition from /etc/magic(.mime).

>   I have not yet fiddled around with thes things but I would regard
> it as very comfortable if I would be able to
>      file <unknown biological data file>
> and get a reasonable and reliable answer.

There are a few differences between the fd.o shared-mime-info database
and the file magic database. The types are IIRC a bit different. The
"offset_start:offset_end" feature of shared-mime-info can be achieved
with the search keyword in file. And file can compute some offsets or
continue at an (computed) offset, which is impossible with the fd.o

My (chemical-mime-data) docs about the different systems are not yet
ready to be given to the community, so you still have to read the docs
of the different systems yourself :) But I have a collection at

HTH & Regards, Daniel

PS: Just ask if you need some assistance. Maybe I can help :)

Reply to: