Missing Hyphens in pdf Export (PDF/A-1a only)

users
#1

I have a fairly large (446 pages) but straightforward LibreOffice Writer
document from which I need to create a PDF/A-1a. When I exported to pdf, I
noticed that none of the hyphens showed up in the pdf; wherever there was
supposed to be a hyphen, there was a space, but no hyphen character.

I "printed" the same document using the CUPS-pdf printer, and the hyphens
appeared fine, but since I require more control over the pdf than that
driver provides, that isn't a good solution.

A little further testing showed that the hyphens disappeared whenever I
chose the PDF/A-1a option, and the problem can be reproduced fairly easily
by creating a document with several justified paragraphs, each using
different fonts if desired. Create two pdf exports - one with PDF/A-1a
checked, and one without it checked.

I'm using LibreOffice Version: 4.1.2.3, Build ID:
40b2d7fde7e8d2d7bc5a449dc65df4d08a7dd38 on 64 bit Ubuntu 12.04 LTS, but have
no idea if this behavior appears in earlier releases, since I never needed
to use the PDF/A-1a option.

Any comments would be welcome.

#2

This is a known bug introduced in 4.1 line:
https://bugs.freedesktop.org/show_bug.cgi?id=67370

It affects all tagged PDF files (PDF/A-1 must be tagged, so is hit as
well) on Linux only.

Therefore, you can workaround this bug by:
- not using tagged PDF
- downgrading to 4.0.6 (or installing it in parallel and using it only
  for exporting PDFs)
- using Windows to create PDFs (in virtual machine, on another machine
  or even trying to run LO through wine - although I don't know if it
  works).

#3

Thanks, Mirosław.

I downloaded LibreOffice into a Windows installation I had running under
Oracle's VirtualBox and, sure enough, the output pdf contained the hyphens
under all conditions.

This seems like really strange behavior; I'd be very curious to know what
coding error could cause such a limited and precise symptom.

For the moment, however, this solves my problem. Is there a way I can find
out when (or in what release) that will be corrected, as it is important to
me.

Again, thanks for the quick response.

Frank

#4

This seems like really strange behavior; I'd be very curious to know
what coding error could cause such a limited and precise symptom.

As bibisect report showed up, bug was introduced between snapshot
3def5194ddaf9c4d766b71527874bd1a973b43e5 and snapshot
9a7603187eb5cc580d33212ee147f9ac89de55f4.

If you can read code, you can take a look at differences between these
two and figure it out:
http://cgit.freedesktop.org/libreoffice/core/diff/?id=9a7603187eb5cc580d33212ee147f9ac89de55f4&id2=3def5194ddaf9c4d766b71527874bd1a973b43e5

For the moment, however, this solves my problem. Is there a way I can
find out when (or in what release) that will be corrected, as it is
important to me.

If you are registered at bugzilla, then go to bug page and in upper
right corner you will find checkbox right next to "Add me to CC list".
Mark that box and hit "Save changes" above.
You will then get e-mail about each change in bug; also when it get
fixed.

#5

Is there a list of Unicode characters that it affects?
U+002D only?

U+002D, U+2010, U+2011, U+2212, U+2013, U+2014, U+2015, U+2E3A,
U+2E3B, U+005F, U+00AD, U+00AF, U+02D7, etc?

jonathon

#6

Is there a list of Unicode characters that it affects?

No, there is not.

U+002D, U+2010, U+2011, U+2212, U+2013, U+2014, U+2015, U+2E3A,
U+2E3B, U+005F, U+00AD, U+00AF, U+02D7, etc?

Well, you can try it yourself and post your finding at mentioned bug
report.
But apparently, only automatically inserted hyphens are affected.

#7

I'm not at all sure if this is related to the same pdf export bug, but I just
noticed that if I have "Small Caps" selected as a font format for any of the
content styles in the table of contents, with a dot leader tab, none of the
dot leaders show up in the pdf output under any conditions at all (i.e. it
doesn't need to be specified as a PDF/A for instance). The dot leaders do
show up normally in the Writer document though.

I've tried this with several fonts and the result is the same. For the
record I'm using LibreWriter Version: 4.1.3.2
Build ID: 70feb7d99726f064edab4605a8ab840c50ec57a on 64 bit Ubuntu 12.04
LTS.