Issue 106523 - Export to PDF/A - 64bit OS
Summary: Export to PDF/A - 64bit OS
Alias: None
Product: General
Classification: Code
Component: code (show other issues)
Version: OOo 2.4.1 RC2
Hardware: PC (x86_64) All
: P3 Trivial (vote)
Target Milestone: OOo 3.2
Assignee: h.ilter
QA Contact: issues@framework
: 98599 108985 (view as issue list)
Depends on:
Blocks: 99999
  Show dependency tree
Reported: 2009-11-01 11:27 UTC by rjanku
Modified: 2017-05-20 10:24 UTC (History)
6 users (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---

how about this (9.12 KB, patch)
2009-11-05 16:58 UTC, caolanm
no flags Details | Diff
that'll do it. icc was also using "long" as the default 32bit type (9.44 KB, patch)
2009-11-05 18:04 UTC, caolanm
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this issue.
Description rjanku 2009-11-01 11:27:57 UTC
PDF document after export from .odt, .ods etc. to PDF/A format have a bug, if is
open in Acrobat Reader (8.x, 9.x). Acroread report: "There was an error
processing a page. Invalid ColorSpace." Export to PDF/A from 3.1.1 in
32-bit OS Fedora (11), Win2k and Win XP is correct.
Comment 1 Olaf Felka 2009-11-02 06:46:47 UTC
@ hi: Please have a look.
Comment 2 caolanm 2009-11-05 16:42:57 UTC
*cough*, take a look a the in icc...

if "$(CPU)"=="I"
... little endian
... big endian

I think I can see where the error is here.
Comment 3 caolanm 2009-11-05 16:57:54 UTC
looks like no-one previously noticed that this doesn't work on any little endian
platforms except for the 32bit intel one.
Comment 4 caolanm 2009-11-05 16:58:21 UTC
Created attachment 65964 [details]
how about this
Comment 5 caolanm 2009-11-05 16:59:35 UTC
I guess we're not running configure and friends in there for a reason, maybe
build time. So how about we merge the big and little endian patch, add a depend
on sal and hack the header to use our own sal endian flags instead ?
Comment 6 philipp.lohmann 2009-11-05 17:17:17 UTC
Ooops, yes that would be a good idea. Will commit to CWS vcl107; if you want
that in 3.2, please convince
Comment 7 caolanm 2009-11-05 17:21:56 UTC
bah, that's only half the story. there's still another bug in the generation of
that for 64bit platforms. I'll have another look to find the other problem
Comment 8 caolanm 2009-11-05 18:04:29 UTC
Created attachment 65967 [details]
that'll do it. icc was also using "long" as the default 32bit type
Comment 9 philipp.lohmann 2009-11-05 18:34:04 UTC
committed in CWS vcl107
Comment 10 philipp.lohmann 2009-11-05 18:47:21 UTC
adjusting title since this affects all 64 bit versions
Comment 11 philipp.lohmann 2009-11-09 14:54:29 UTC
*** Issue 98599 has been marked as a duplicate of this issue. ***
Comment 12 strata 2009-11-10 00:04:13 UTC
I just have posted this now fixed issue in the similar (?!) one:

for which I have been a good tester all over the months.


p.s. Intelligent proposal/suggestion: why doesn't work to win32,
win64 (!!!) and MacOSX develop and doesn't work on the linux/*BSD/...
develop, does it?!!! And both communities ties together with an active and open
minded ideas and code exchanges?!!! 
Comment 13 rene 2009-11-10 13:28:31 UTC
I more think is the right
bug, not 440055
Comment 14 rene 2009-11-10 13:54:21 UTC
Comment 15 Martin Hollmichel 2009-11-10 14:31:56 UTC
isn't it quite ugly to make OOo's sal a prerequisite for building external icc ?

I would appreciate a more generic patch to avoid unwanted dependencies?
Comment 16 philipp.lohmann 2009-11-10 14:38:28 UTC
I'd suggest postponing purely cosmetical changes until the very real problem is
fixed. If cmc wants to do a sal typeless version, it would of course be welcome,
but I think this is a very good fix as it is.
Comment 17 uwe.luebbers 2009-11-10 16:30:19 UTC
adjusted target
Comment 18 philipp.lohmann 2009-11-10 18:27:13 UTC
now also committed to CWS ooo32gsl05 for 3.2 target
Comment 19 philipp.lohmann 2009-11-12 12:27:32 UTC
please verify in CWS ooo32gsl05
Comment 20 h.ilter 2009-11-17 11:18:02 UTC
Verified on PL's machine and compared between master & CWS = OK
Comment 21 rjanku 2009-11-19 11:37:01 UTC
Big thank for fix, adjusting target and verification to all solvers. And thanks
to the Fedora developers for quickly apply - after today update OOo in Fedora 11
64-bit is problem solved.
Comment 22 philipp.lohmann 2010-02-05 13:52:43 UTC
*** Issue 108985 has been marked as a duplicate of this issue. ***