acroread 9.5.5 (linux) shows binary garbadge after page numbers

[Henri Menke]

On Fri, 2021-08-20 at 11:45 +0200, Harald Koenig via tex-live wrote:
> Hi TL team,
> 
> I just notice that with TL-2021 pdflatex, I see binary garbage after
> page numners (see upper left in attached screenshot).
> 
> looking into the PDFs generated by TL-20 and TL-21 I see:
> 
> ---8<---    TL-20  ---8<------8<------8<------8<------8<---
>   /PageLabels <<
>     /Nums [
>       0
>       <<
>         /P (1)
>       >>
>       1
>       <<
>         /P (2)
>       >>
> ---8<---    TL-21  ---8<------8<------8<------8<------8<---
> /PageLabels <<
>     /Nums [
>       0
>       <<
>         /P <feff0031>
>       >>
>       1
>       <<
>         /P <feff0032>
>       >>
> ---8<------8<------8<------8<------8<------8<---
> 
> 
> so the new byte order mark are part of that problem (and of course
> crappy old and broken acroread for linux:-(

Adobe Reader for Linux is 32-bit only and has tons of unfixed
vulnerabilities including remote code execution.

https://www.cvedetails.com/product/497/Adobe-Acrobat-Reader.html?vendor_id=53

We can now also add broken Unicode support to the list of problems.

Seriously, you'll have to find another viewer.

> I also find byte order marks in some but not all of the PDF metadata
> (and only there and page numbers):
> 
> 72 0 obj
> <<
>   /Author
> <feff00200048006100720061006c00640020004b00f6006e006900670020>
>   /CreationDate (D:20210820105934+02'00')
>   /Creator
> <feff004c0061005400650058002000770069007400680020004200650061006d006500
> 7200200063006c006100730073>
>   /Keywords ()
>   /ModDate (D:20210820105934+02'00')
>   /PTEX.Fullbanner (This is pdfTeX, Version 3.141592653-2.6-1.40.23
> \(TeX Live 2021\) kpathsea version 6.3.3)
>   /Producer (pdfTeX-1.40.23)
>   /Subject ()
>   /Title <feff005300480045004...
>   /Trapped /False
> > > 
> endobj
> 
> 
> "fun" fact aside: acroread correctly displays the Author and Title in
> document properties:-(
> 
> 
> so my more generic questions:
> 
> - why are byte order marks inserted for some trival 7-bit-ASCII strings
> like page numbers (1) (2) (3) etc ?
>   here they're not needed for sure -- author names and titles are a
> different story of course...
>   
> - why do some PDF metadata not have byte order marks ?
> 
> - is this a problem for other PDF tools/viewers too ?
> 
> 
> 
> thanks for any comments (other than "don't use acroread" ;-)
> 
> at least for now I can continue using TL-20,
> and I think about writing a small PDF editing/hacking script using
> 
>     sed 's/ <feff/ </g'
> 
> ;-)
> 
> 
> Harald