Re: apsfilter 5.4.2 bugs'n'stuff

[Matej Vela <vela@xxxxxxxxxx>]

On Tue, May 09, 2000 at 11:45:05AM +0200, Michael Loin wrote:
> Matej Vela wrote:
> > On Tue, May 02, 2000 at 12:30:53PM +0200, Michael Lossin wrote:
> > > * For a local printer, $PRINT_REDIR is not set, so print_data
> > > and print_ascii_recode end in void (which most probably is the
> > > wrong destination). It should be:
> > > [...]
> > > I don't know what should happen in print_data... any hints?
> > 
> > I'm sorry, I didn't understand this?
> 
> What I meant was: should the pipe end in "print_raw" or just
> "$PRINT_REDIR"? (print_raw should be alright, I guess...)

Data (i.e. PCL, PJL, etc) should already be in a format suitable for
printing, so I think that it can go directly to $PRINT_REDIR.  (If it
isn't printable, print_raw won't help.)

(...)
> * Why is apsfilter relying on the filters_found file? Wouldn't it be
> simpler to check for the programs at run-time?
> (change  'if [ -n "$HAVE_xyz" ]' to 'if which >/dev/null xyz')
> Of course it would be uglier, but WTF... No need to re-run the filter
> setup every time.

Using which (even if we implement it as a shell function) imposes a
performance penalty.  However, since it does make our users' lives
simpler, I think we can live with that.

I'll wait for Andreas to see whether he agrees.

> * (CVS version only) Using acroread in a pipeline results in a /tmp file
> left after the script has ended (the "-pairs" construction with two
> temporary files is needed).

(Thanks for testing it!)

Should we bother to work around this?  It will probably be fixed soon,
and /tmp is cleaned automatically anyway.

If you really care, try the attached acroread wrapper -- it fixes the
problem for everybody, and not just apsfilter.

> * I suggested the use of "convert" for various bitmap (and other)
> formats. Has anyone done serious comparison of the output quality to
> the one generated by "...topnm | pnmtops"? (IMHO it really depends on
> the input file.)

Where can I download "convert"?

> * 'This may uglify the source' (TM), but mpage is a tool similar to a2ps,
> wouldn't it be nice... *ouch* Don't hit me! :)

I've added preliminary support.  Please test.

Thanks,

Matej
#!/bin/sh -e
# (C) 2000  Matej Vela <vela@debian.org>
# Under the GNU GPL, version 2 or higher.
#
# Move acroread to acroread.real, and put this script in its place.

no_pairs=true
topostscript=false
for argument; do
	case "$argument" in
		-pairs)		no_pairs=false		;;
		-toPostScript)	topostscript=true	;;
	esac
done

if $no_pairs && $topostscript; then
	: ${TMPDIR:=/tmp}
	pdf_file="$TMPDIR/acroread$$.pdf"
	ps_file="$TMPDIR/acroread$$.ps"

	umask 077
	set -C
	> "$pdf_file"
	> "$ps_file"
	set +C

	set -- "$@" -pairs "$pdf_file" "$ps_file"
fi

acroread.real "$@"

if $no_pairs && $topostscript; then
	rm "$pdf_file" "$ps_file"
fi

exit 0