Re: Breaking programs because a not yet implemented solution exists in theory
On Wed, Feb 01, 2012 at 01:51:17AM -0800, Russ Allbery wrote:
> Josselin Mouette <email@example.com> writes:
I confirm that I agree that we should prevent duplication of data which
was stated in previous mails.
> The main place that mailcap is richer than the desktop file that I can see
> is that mailcap allows you to express the exact command line (including
> putting %s at different places if needed) and lets you specify different
> commands for viewing, editing, and printing. For example:
> message/rfc822; mutt -Rf '%s'; edit=mutt -f '%s'; needsterminal
> I don't know if there's any way to do that with desktop files.
> Also, I don't think there's a desktop equivalent of copiousoutput.
Assuming that Russ did not overlooked something this means that mailcap
entries can not generatet from desktop files. So the one-liners
mentioned by Josselin which might solve 50% of the task could not easily
enhanced to two-liners doing 100% which do all the work. In other words
we dropped support for a technique that is used by several programs and
it seems a replacement is either hard to do or not possible at all.
I personally would cope with this by installing a local package carrying
the mailcap entries I need. However that can hardly be a solution for
our users. As a general solution I would see two ways:
1. Stop droping *.mime files from packages and reinjecting them.
2. Create a general mailcap entry collection package which works
around maintainers unwilling to support mailcap.
I'd prefer 1. because I see no point in just droping what worked in the
past and has no visible chance to break something heavily. Please
correct me if I'm wrong.