Registering experimental mime types

[Bastien Nocera]

On Tue, 2010-09-07 at 01:22 +0200, Giovanni Campagna wrote:
> (warning: this is really long)
> 
> Studying Freedesktop's shared-mime-info, I noticed a lot of x- mime
> types, which according to RFC4288 (Media Type Specifications and
> Registration Procedures), section 3.4
> 
> "These types are unregistered, experimental, and for use only with the
> active agreement of the parties exchanging them."
> 
> should not be used in a standard specification, as they cannot be
> interoperable. In particular, it means that Web or Mail applications
> cannot interoperate safely with the shared mime database, as HTTP and
> MIME don't follow Freedesktop recommendations (but do follow IETF).
> 
> There are many of these, but some are more important, because they
> represent types directly under Freedesktop responsibility and thus
> should be formally registered with IANA. Also, I noted that for some
> weird reasons, many types have a non experimental alias listed within
> them. The database should be changed to expose the official media type,
> instead of the x- type.

We know most of those rules already...

> So, let's show the result of reading through the database. First of all,
> the primary media types (before the slash, that is), which require an
> RFC:
> 
> - x-content/*: as it is unlikely that IANA will allow registering
> content/*, it may require renaming to unknown-content/*,
> generic-content/*, or better to media-support/*, since that is what it
> is used for currently

x-content/* is a freedesktop extension to allow "mime-types" to be
associated with mounts. I don't see what's wrong with prefixing it with
x-, noting that it's a non-standard content-type.

> (as a side note: unix is a registered trademark)

What does that have to do with the above?

> - inode/*: this does not use x-, but it is not a registered media type,
> according to http://www.iana.org/assignments/media-types/index.html -
> registration is necessary, I think

Most of those are left-overs from the gnome-mime-data times. Feel free
to come up with "more correct" ones.

> - font/*: I'm not sure font faces can be stuffed inside application/*, a
> new primary media type may be discussed

shared-mime-info doesn't ship with those.

> Secondly, I believe the following need full RFC standarization, as their
> format is widely used and possibly already a recognized standard
> 
> - application/x-executable => application/executable
> - application/x-core => application/core-file
> - application/x-object => application/object-file
> - application/x-sharedlib => application/shared-library
> - application/x-bzip => application/bzip
> - application/x-gzip => application/gzip
> - application/x-lzma => application/lzma
> - application/x-xz => application/xz
> - application/x-tar => application/tar-archive
> - application/x-bzip-compressed-tar => application/tar-archive+bzip
> - application/x-compressed-tar => application/tar-archive+gzip
> - application/x-lzma-compressed-tar => application/tar-archive+lzma
> - application/x-xz => application/tar-archive+xz
> - application/x-cd-image => application/iso-cd-image
> - application/x-matroska => application/matroska
> - video/x-matroska => video/matroska
> - audio/x-matroska => audio/matroska
> - audio/x-vorbis+ogg => audio/vorbis+ogg
> - audio/x-flac+ogg => audio/flac+ogg
> - audio/x-speex+ogg => audio/speex+ogg
> - audio/x-speex => audio/speex
> - video/x-theora+ogg => video/theora+ogg
> - video/x-ogm+ogg => video/ogm+ogg
> - application/x-pkcs7-certificates => application/pkcs7-certificates
> - application/x-pkcs12 => application/pkcs12
> - application/x-x509-ca-cert => application/x509-ca-cert
> - text/x-adasrc => text/source-code-ada
> - text/x-c++hdr => text/source-code-cpp-header (I think + is a reserved
> char)
> - text/x-c++src => text/source-code-cpp
> - text/x-chdr => text/source-code-c-header
> - text/x-csrc => text/source-code-c
> - text/x-csharp => text/source-code-csharp
> - text/x-fortran => text/source-code-fortran
> - text/x-pascal => text/source-code-pascal
> - text/x-sql => text/source-code-sql

Do you have RFC references for all those?

> Then, various freedesktop specific x- subtypes should go in the
> freedesktop vendor subtree.
> - application/x-xbel => application/vnd.freedesktop.xbel+xml
> - application/x-desktop => application/vnd.freedesktop.desktop-entry
> - application/x-theme => application/vnd.freedesktop.theme
> - application/x-trash => application/vnd.freedesktop.backup-file
> - application/x-zerosize => application/vnd.freedesktop.empty
> - image/x-xcursor => image/vnd.freedesktop.cursor (or
> image/vnd.xorg.cursor)
> - image/x-xbitmap => image/vnd.xorg.bitmap
> - image/x-xpixmap => image/vnd.xorg.pixmap

Using x- might be antiquated, but it's still valid.

> Other free software vendor specific media types (for vendors likely to
> follow the xdg list)
> - application/x-abiword => application/vnd.abiword.document
> - application/x-dia-diagram => application/vnd.gnu.dia-diagram
> - application/x-dia-shape => application/vnd.gnu.dia-shape
> - application/x-deb => application/vnd.debian.package
> - application/x-designer => application/vnd.nokia.qt-designer
> - application/x-e-theme => application/vnd.enlightment.theme
> - application/x-gettext-translation =>
> application/vnd.gnu.gettext-machine-object
> - application/x-glade => application/vnd.gnome.glade
> - application/x-gnucash => application/vnd.gnu.cash
> - application/x-gnumeric => application/vnd.gnu.spreadsheet
> - application/x-gnuplot => application/vnd.gnu.plot
> - application/x-karbon => application/vnd.kde.karbon
> - application/x-kchart => application/vnd.kde.kchart
> - application/x-kexi-connectiondata =>
> application/vnd.kde.kexi-connectiondata
> - application/x-kexiproject-shortcut =>
> application/vnd.kde.kexi-shortcut
> - application/x-kexiproject-sqlite? => application/vnd.kde.kexi-sqlite?
> - application/x-kformula => application/vnd.kde.kformula
> - application/x-killustrator => application/vnd.kde.killustrator
> - application/x-kivio => application/vnd.kde.kivio
> - application/x-kontour => application/vnd.kde.kontour
> - application/x-kpovmodeler => application/vnd.kde.kpovmodeler
> - application/x-kpresenter => application/vnd.kde.kpresenter
> - application/x-krita => application/vnd.kde.krita
> - application/x-kspread => application/vnd.kde.kspread
> - application/x-kspread-crypt => application/vnd.kde.kspread-crypt
> - application/x-ksysv-package => application/vnd.kde.ksysv-package
> - application/x-kugar => application/vnd.kde.kugar
> - application/x-kword => application/vnd.kde.kword
> - application/x-kword-crypt => application/vnd.kde.kword-crypt
> - application/x-mozilla-bookmarks => application/vnd.mozilla.bookmarks
> - application/x-nautilus-link => application/vnd.gnome.nautilus-link
> - application/x-oleo => application/vnd.gnu.oleo
> - application/x-profile => application/vnd.gnu.profile
> - application/x-rpm => application/vnd.redhat.package
> - application/x-shared-library-la => application/vnd.gnu.libtool-archive
> - image/x-compressed-xcf => image/vnd.gnu.compressed-xcf
> - image/x-xcf => image/vnd.gnu.xcf
> - message/x-gnu-rmail => message/vnd.gnu.rmail
> - text/x-vala => text/vnd.gnome.source-code-vala
> - text/x-emacs-lisp => text/vnd.gnu.emacs-lisp
> - text/x-gettext-translation => text/vnd.gnu.gettext
> - text/x-gettext-translation-template => text/vnd.gnu.gettext-template
> - text/x-moc => text/vnd.nokia.qt-moc
> - text/x-rpm-spec => text/vnd.redhat.package-spec
> - application/x-xpinstall => application/vnd.mozilla.xpinstall
> 
> (I marked every application/x-k* as part of the KDE project, some may
> actually be independent apps for the KDE platform. Nevertheless, there
> are a lot of k* mime types, it would be good for those apps to go for
> standard defined formats)
> 
> Then the standard Unix file types (beyond those covered by inode/*).
> In this case, and if this proposal is approved, the Open Group should be
> asked for a formal registration.
> - application/x-archive => application/vnd.unix.ar
> - application/x-awk => application/vnd.unix.awk
> - application/x-bcpio => application/vnd.unix.binary-cpio
> - application/x-compress => application/vnd.unix.compress
> - application/x-cpio => application/vnd.unix.cpio
> - application/x-m4 => application/vnd.unix.m4
> - application/x-reject => application/vnd.unix.patch-reject
> - application/x-shar => application/vnd.unix.shell-archive
> - application/x-shellscript => application/vnd.unix.shell-script
> - application/x-tarz => application/tar-archive+vnd.unix.compress (is
> this legal?)
> - application/x-troff-man => application/vnd.unix.manual-page
> - application/x-troff-man-compressed =>
> application/vnd.unix.manual-page-compressed
> - text/x-makefile => text/vnd.unix.makefile
> - text/x-patch => text/vnd.unix.patch
> 
> I'm not sure about the following
> - text/x-authors
> - text/x-changelog (=> text/vnd.gnu.changelog ?)
> - text/x-copying (=> text/vnd.gnu.license ?)
> - text/x-credits
> - text/x-install
> - text/x-log
> - text/x-readme
> - text/x-uri (=> text/vnd.freedesktop.uri ?)
> 
> A lot more experimental media types exist, but cannot be changed without
> cooperation with the responsible organization, which should be
> contacted. I have a full list if someone wants it.
> 
> I believe this change will help the shared-mime-info becoming the
> de-facto standard for converting from a file (with a name and some data)
> to a full media type, helping the adoption of Freedesktop and Internet
> standards.
> 
> I know it will require a lot of work, not only changing the database and
> sending mails to IANA, but also updating the applications accordingly,
> but I think it is worth it.

If you're willing to do the legwork getting those new mime-types
accepted with the IANA, I'd be happy for you to do that. In the
meanwhile, I don't see many reasons to change our current way of doing
things.

Cheers