Closed GoogleCodeExporter closed 8 years ago
Is there a test that is failing ? How do I make sure this solve an issue ? How
do I make sure I wont create a regression after applying this patch ? thx
Original comment by mathieu.malaterre
on 24 Feb 2014 at 8:44
There can not be a regression.
If 'WORDS_BIGENDIAN' is used, these patches allow to use 'WORDS_BIGENDIAN
at all relevant places.
The additional correction in line 69 of cio.c replaces '4' with
'sizeof(OPJ_UINT32)'.
The correction in line 47 of cio.c is necessary for BE; otherwise
the code is the same as in LE.
winfried
Original comment by szukw...@gmail.com
on 24 Feb 2014 at 5:50
Unfortunately I am no longer sure about the code in line 47 of cio.c : you have
to
wait for 10 days.
winfried
Original comment by szukw...@arcor.de
on 24 Feb 2014 at 6:56
Mathieu,
here is the ENDIAN patch for openjpeg-trunk-r2578. I use 'WORDS_BIGENDIAN'.
If you prefer 'OPJ_BIG_ENDIAN' ( Why? ), you must replace ALL occurrences
of 'WORDS_BIGENDIAN' with 'OPJ_BIG_ENDIAN'; even in the binaries.
The test
#if WORDS_BIGENDIAN == 0
is very old. It can be used only with
#cmakedefine01 WORDS_BIGENDIAN
This patch has been tested with a G5 machine.
winfried
Original comment by szukw...@arcor.de
on 6 Mar 2014 at 1:38
Attachments:
Original comment by mathieu.malaterre
on 19 Mar 2014 at 1:15
I prefer OPJ_BIG_ENDIAN because it does not pollute the global namespace, it
uses an OPJ_ prefix. OPJ_BIG_ENDIAN is used in openjpeg 2.0 anyway. This makes
it easier for me to backport stuff to 1.x branch.
Original comment by mathieu.malaterre
on 25 Mar 2014 at 4:46
I am tempted to close this as a dup of issue 302
Original comment by mathieu.malaterre
on 26 Mar 2014 at 3:28
Original issue reported on code.google.com by
szukw...@arcor.de
on 28 Sep 2013 at 3:59Attachments: