Alexander Zangerl writes:
> On Sun, 08 Nov 2015 15:18:22 +0100, Harald Geyer writes:
> >"run-mailcap --action=view" is meant to start a pager, which is not
> >what mhshow wants to do.
>
> why not? where does it say that?
I don't know if this is written down anywhere, but it can be infered
from the fact that mhshow does start a pager of its own.
> view is meant to do whatever is necessary to present the material to
> the user.
Agreed.
> >mhshow wants formatted output that can be
> >passed on to it's own pager (moreproc).
>
> not necessarily; look at the examples in man mhshow for, say, audio
> or image/*:
>
> mhshow-show-audio/basic: raw2audio 2>/dev/null | play
> mhshow-show-image: xv %f
> mhshow-show-application/PostScript: lpr -Pps
Well, these of course can not be displayed in a pager in the first place ...
> there is no paging by nmh involved in any of these. there is nothing
> in the mhshow docs that indicates that a pager supplied by mhshow must
> be used.
I haven't read the source of mhshow, but from what I have seen with
strace (yes, actually needed strace to figure out what's wrong) it looks
to me like mhshow always fires up moreproc to display the headers *and*
any text/* parts after they have been filtered by the respective command
from mhn.defaults - I think this implies that mhshow expects the entries
in mhn.defaults to behave as filters - at least if X is not available.
> >x/console/whatever use, but nmh (and sensible-tools) are messing
> >things up, because they try to do part of the job themselfs. I think
> >the easy way to fix this is to dump run-mailcap down so it doesn't
> >interfere with what nmh is doing.
>
> to be honest i have not yet seen any of the interference problems
> you're having. with the default mhn.defaults, no special .mh_profile entries
> whatsoever, and a pure html email (yuck), for me mhshow
>
> a) fires up my real browser if i have a $DISPLAY,
Yes.
> b) falls back to showing me the transmogrified text (via w3m)
> and paged by less if there is no $DISPLAY.
Looks like w3m is smarter than links2 and doing the right thing
when stdout is not a terminal.
> if i switch the mhn.defaults to --action=cat, then case a) never
> happens anymore, because action cat ignores all mailcap entries but the
> ones with 'copiousoutput' and not many have that set,
Yes, but I have proposed a solution to that in my original message,
in case you acutally think this is desired.
> and in the remaining case b) no paging whatsoever is performed.
This is not true. The paging is done by moreproc (less on my system).
If this is not the case on your system, then I see where your opposition
is coming from. However I belive something else must be broken in that
case.
> i find this latter setup highly undesirable.
>
> >I think your explanation in the News.Debian is a good one.
>
> ok, then i'll expand that in the next upload to mention that people
> should also consider run-mailcap --action=cat as alternative to
> the defaults...
>
> >configuration probably works on many sytems - but only by chance, not
> >by being right.
>
> ...because i disagree with your opinion that there is
> anything wrong with delegating show actions to run-mailcap --action=view.
I think we are not yet disagreeing about opinion, we are still disagreeing
about facts (ie when is moreproc fired up). Let's first sort out the
facts, then move on to opinions.
> so, shall we agree to disagree? i just don't think there's a single
> setup that'll match everybody's preferences.
I believe this is not about my preferences (that's what config files
are for) but about software compatibility. And software compatibility
is a matter of debian packaging. Obviously I had a bad start at
explaining what the issue is - it's your decision if the issue is worth
fixing (ie tag the bug report wontfix), but I want you to actually see
the issue first. (Once we are there, maybe you can tell me how I could
have presented the issue better.)
HTE,
Harald
