On Tue, Nov 02, 2021 at 04:10:44PM +0100, Jan Stary wrote:
> This is current/amd64 on a PC, using lpr with this /etc/printcap:
> lp::lp=:rm=pr.stare.cz:rp=lp:sd=/var/spool/output/lpd:lf=/var/log/lpd-errs:sh:
> which is a Brother DCP9055CDN via ethernet.
>
> Now, I have this pdf file (attached), broken in a way that puzzles me.
> When viewed with mupdf or gv, it shows one thing, when printed with lpr,
> it shows something else.
>
> I don't mean a missing glyph when an exotic language is used; it is
> in Czech, but it's not the rendering of Czech letters that's strange:
> it has 11 printed instead of 10 (line 1), 22 instead of 29 (line 2),
> 7000 instead of 7084 and 5222 instead of 5268 (line 4),
> 3 h 33 mm instead of 3 h 30 min (line 5.1), etc.
>
> Agonizing moments have been spent looking at the page,
> making sure it is actualy the file.
>
> Vaguely speaking, in these examples, some chars/glyphs
> seem to be repeated in the print, instead of the next one
> that should have been printed:
>
> 11 not 10 (repeating the 1)
> 22 not 29 (repeating the 2)
> 7000 not 7084 (repeating the 0)
> 33 mm not 30 min (repeating the 3 and the m)
>
> Is that an indication of some particular kind
> of breakage in a pdf file?
>
> Inside the pdf, I see
>
> </Producer(GPL Ghostscript 8.15)
> /CreationDate(D:20150306075816)
> /ModDate(D:20150306075816)
> /Title(klic_5_tridy.xlsx)
> /Creator(PScript5.dll Version 5.2.2
>
> so I suppose the file was produced as a pdf export of
> a xlsx file by some awful office package or another,
> probaly on windows (dll).
>
> When printed from gv, it prints what gv and mupdf show.
> When printed at a corporate myq print system, it prints the same;
> but when printed with lpr, it prints these strange alterations.
>
> I don't think it's lpr's fault, so this might not even be the list,
> for lpr just sends what it gets (except wrapping it in the cf, df files
> of the lpr protocol, right?), but I would still like to know:
> is it that gv can somehow interpret the broken pdf in the right way,
> sending the right bits to the printer to print, but the Brother printer
> (i.e., the printer's pdf interpreter?) can not?
> Please excuse my pdf/ps ignorance.
>
> gv also says
>
> Warning: Missing charsets in String to FontSet conversion
>
> when viewing the file, but not with LC_CTYPE=C;
> normally my env has LC_CTYPE=en_US.UTF-8.
> There are some Czech letters printed wrong,
> but surely every FontSet (whatever that is)
> has glyphs for the digits.
>
> Thank you
>
> Jan
>
Hi Jan
FWIW, I cannot reproduce this on my Brother HL-L3270CDW with lpr, file
prints fine. This is on a -current (well, current-ish, October 5) amd64
machine, without cups, foomatic or anything like that.
/etc/printcap:
lp:\
:lp=:rm=10.17.19.134:\
:sh:\
:lf=/var/log/lpd-errs:
--
