On 06.08.2012 12:47, Jürgen Schmidt wrote:
On 8/6/12 12:30 PM, Rory O'Farrell wrote:
On Mon, 6 Aug 2012 07:10:00 -0300
Ariel Constenla-Haile <[email protected]> wrote:
Hi Jürgen, *
On Mon, Aug 06, 2012 at 11:36:01AM +0200, Jürgen Schmidt wrote:
Hi,
we have currently 2 issues that need further investigation
- print problem with multiple pages
https://issues.apache.org/ooo/show_bug.cgi?id=120464
I can't reproduce it on MacOS but reproducable on Windows and others
confirmed Linux as well
- documents with password can't be loaded. This seems to be Linux only
problem because we can't reproduce it on MacOS and Windows (clean builds
on rev. 1369110)
I assume it's a 64 bits only issue (it happens that we support 64 bits
only in Linux), I base my assumption in the following:
- I can't reproduce it on Ubuntu 10.4 32 bits
- I can reproduce it on
- Ubuntu 10.4 64 bits
- Fedora 17 64 bits
- Josef reproduced it on Ubuntu 12.04 64 bits
- Rory couldn't reproduce it on Ubuntu 11.10 32 bits
This can't be a build bug: I made a full clean build (and just for
testing once again, I rebuilt it just now, and is still reproducible).
I've just installed rev. 1369110 on a dual boot machine (Ubuntu 11.10 32 bit and Windows
XP 32 bit). Working to the same printer, Ubuntu printed the four pages of the test file
correctly, Windows printed the first page, followed by three blank pages (correct page
count). I tried Windows printing to page 1-4, with same result. But if I tell it to
print in reverse order, it prints all pages! It will also print any page selected by
number. Having done all that, it still will only print page 1 followed by three blank
pages if I tell it to print "all".
So Ubuntu 11:10 32bit passes
Windows XP 32 bit fails
build 9110 contains no fix for both described issues above.
- the print problem is analyzed and already fixed, it will be in the
next snapshot
- the password problem is under investigation ...
64 bit only problem sounds reasonable but we have to check what's going
wrong.
This really is a 64bit-only problem. Due to a recent modification
encryption and digest type are internally stored as 64bit values on
64bit compilers. When they are later converted into 32bit values the
conversion fails silently and the verification of the password fails also.
With 32bit compilers the values are stored internally as 32bit values
and the conversion is OK.
For details see issue 120385
(https://issues.apache.org/ooo/show_bug.cgi?id=120385).
Fixed in SVN revision 1369843 on branch AOO34.
-Andre
