[Cin] OpenEXR 2.4 - try3!

Andrew Randrianasulu randrianasulu at gmail.com
Thu Mar 19 05:32:37 CET 2020 

В сообщении от Thursday 19 March 2020 06:30:14 Phyllis Smith написал(а):
> This OpenEXR package is a "bag of worms" -- weird, hard, and worse package
> to integrate in Cinelerra.  GG has taken your patches, made quite a few
> changes, and got it to work at 2.4.1.  Tomorrow he will test to see how it
> does on the 18+ distros to see if it will work for the monthly builds.

If it break - I can try to search for possible solutions. I saw CentOS 
mentioned in documentation update - depend on version this may require some 
thinking and action.

After all - I made this motion - therefore I should help if it broke. And yes, I tried to push 
this rel. early because doing this stuff at near rebuild date is .. more nerve-wrenching.

Still, if it really broke more than one or two builds - I will not push upgrade 
for the sake of upgrade - after all, there are system libraries, and I can patch stuff locally ....

> Meanwhile, I think he has come to the conclusion that it is not worthwhile
> to upgrade it any more because there do not seem to be any gains in doing
> so -- there releases seem to be only code cleanup but not no features
> whatsoever.

Well, if there will be more use of OpenEXR/ILMbase features/functions in Cin - 
then outdated lib will make life harder .... (I'm thinking about those 16-bit float functions)

> 
> Will see if it works for the monthly builds, and then probably never again
> will be upgraded.  gg/phyllis
> 
> On Wed, Mar 18, 2020 at 9:35 AM Andrew Randrianasulu <
> randrianasulu at gmail.com> wrote:
> 
> >
> >
> > В сообщении от Wednesday 18 March 2020 18:00:06 Good Guy написал(а):
> > > These changes look good.   Thanks.
> > > and now the inevitable "but"..., these are just suggestions,
> > > but they are based on my past experiences.
> > >
> > > 1) don't leave dead code in the product unless you think it may be useful
> > > in the future.
> >
> > Well, removed ilmbase from both configure.ac and thirparty/Makefile
> >
> > If you decide -stdc+11 requirement is NOT show stopper, then you can try
> > to apply those
> > and two other patches, also attached this time, because without them
> > nothing will work
> >
> > put openexr-2.4.1 patches into thirdparty/src, put 2.4.1 tarball there,
> > too (you may want to repack it,
> > or I can find xz version ...):
> >
> > https://sources.voidlinux.org/openexr-2.4.1/
> >
> >  sha256sum openexr-2.4.1.tar.gz
> > 3ebbe9a8e67edb4a25890b98c598e9fe23b10f96d1416d6a3ff0732e99d001c1
> > openexr-2.4.1.tar.gz
> >
> > matches with sha256sum provideed at this page:
> > https://www.freshports.org/graphics/openexr
> >
> > Try on oldest system first? (due to c++11 req.  it may fail first).
> >
> > > examples:  the configure script is already sort of trashy, since it is
> > the
> > > driver for a set
> > > of old products.  The main reason the thirdparty design exists is that
> > you
> > > can't trust the
> > > maintainers to not break your build.  PLEASE be aware that every month
> > any
> > > mod you
> > > make has to work on 18 platforms that cover a span of 10 years or so.
> > Any
> > > break that
> > > is historical can stop several builds.  The build system is a xen host
> > > running an
> > > embedded busybox system that only operates the build.  No servers, no
> > > installed extras.
> > > It operates 3 sets of build iterators.  If any of them fail, more passes
> > > (4,5,...).  Each pass
> > > is risky, since it is almost always some kind of emergency patch, and
> > > usually
> > > downgrades the results.  Another example, /usr/X11R7/lib has not been
> > > around for a
> > > while on most platforms.  It also as permuted variants /usr/lib/X11
> > > /usr/share/X11/ etc.
> > > If the change is really needed for a particular platform, then it may
> > need
> > > to be "specialized".
> > >
> > > 2) I am not sure, but it looks like there is a new version of OpenEXR
> > > (2.4.1).  I tried it,
> > > but it requires the a very recent version of cmake.  Currently, there
> > seems
> > > to be a struggle
> > > to replace autotools with cmake,meson,waf ... Meson broke many builds.  I
> > > rewrote the
> > > dav1d meson build as a makefile, and tried to give it to the developers,
> > > but they declined to
> > > carry it.   Every time they tweak the build, I have to mod the makefile.
> > > Not nice.  And now
> > > the suggestion,  If you are going to refactor the OpenEXR build, please
> > see
> > > if it is possbile
> > > to get it to work with the more recent version.   It is not essential to
> > > make it work, since it
> > > is already working with a usable version, but keeping up with 40 or so
> > > libraries is a major
> > > task, and maintenance is continuous.  Try to make sure that the work you
> > do
> > > will be good
> > > for as long as possible.
> > >
> > > 3) I personally wish to thank you (and all of you) who have helped to
> > > improve the result.
> > > It has always been a curiosity that almost all of the technical help is
> > not
> > > from any "techy"
> > > component, (eg. google, apple, adobe, ...) but from real users that are
> > > just plain interested
> > > in having something nice to use.  It is highly appreciated for your work
> > on
> > > a large section
> > > of source code/libraries. Thank you.
> > >
> > > gg
> > > PS.  The documentation rework is just outstanding.
> > >
> > >
> > >
> > >
> > >
> > > On Wed, Mar 18, 2020 at 6:07 AM Andrew Randrianasulu <
> > > randrianasulu at gmail.com> wrote:
> > >
> > > > Only configure.ac patch this time ..
> > > >
> > > > diff --git a/cinelerra-5.1/configure.ac b/cinelerra-5.1/configure.ac
> > > > index f10af67..7375fa9 100644
> > > > --- a/cinelerra-5.1/configure.ac
> > > > +++ b/cinelerra-5.1/configure.ac
> > > > @@ -8,7 +8,7 @@ AC_LANG_CPLUSPLUS
> > > >  AC_LANG_C
> > > >  AC_PROG_CXX
> > > >
> > > > -CFG_CFLAGS+=" -fno-omit-frame-pointer -fno-math-errno
> > -fno-signed-zeros"
> > > > +CFG_CFLAGS+=" -std=c++11 -fno-omit-frame-pointer -fno-math-errno
> > > > -fno-signed-zeros"
> > > >  CFG_CFLAGS+=" -pthread -Wall"
> > > >  # misguided pedantic warnings
> > > >  #  this is extra work, not a gain...
> > > > @@ -239,14 +239,14 @@ PKG_3RD([giflib],[yes],
> > > >      libutil.a ],
> > > >    [ . ])
> > > >
> > > > -PKG_3RD([ilmbase],[auto],
> > > > -  [ilmbase-2.2.1],
> > > > -  [ Iex/.libs/libIex.a \
> > > > -    IexMath/.libs/libIexMath.a \
> > > > -    Half/.libs/libHalf.a \
> > > > -    Imath/.libs/libImath.a \
> > > > -    IlmThread/.libs/libIlmThread.a ],
> > > > -  [ Iex Half Imath config IlmThread ])
> > > > +#PKG_3RD([ilmbase],[auto],
> > > > +#  [ilmbase-2.2.1],
> > > > +#  [ Iex/.libs/libIex.a \
> > > > +#    IexMath/.libs/libIexMath.a \
> > > > +#    Half/.libs/libHalf.a \
> > > > +#    Imath/.libs/libImath.a \
> > > > +#    IlmThread/.libs/libIlmThread.a ],
> > > > +#  [ Iex Half Imath config IlmThread ])
> > > >
> > > >  PKG_DEF([ladspa], [ladspa-0.4.17], [], [])
> > > >
> > > > @@ -337,11 +337,16 @@ PKG_3RD([mjpegtools],[yes],
> > > >      mpeg2enc/.libs/libmpeg2encpp.a ],
> > > >    [ .  lavtools utils ])
> > > >
> > > > -PKG_3RD([openexr],[auto],
> > > > -  [openexr-2.2.1],
> > > > -  [ IlmImf/.libs/libIlmImf.a \
> > > > -    IlmImfUtil/.libs/libIlmImfUtil.a ],
> > > > -  [ IlmImf config ])
> > > > +PKG_3RD([openexr],[yes],
> > > > +  [openexr-2.4.1],
> > > > +    [ IlmBase/Iex/.libs/libIex.a \
> > > > +    IlmBase/IexMath/.libs/libIexMath.a \
> > > > +    IlmBase/Half/.libs/libHalf.a \
> > > > +    IlmBase/Imath/.libs/libImath.a \
> > > > +    IlmBase/IlmThread/.libs/libIlmThread.a
> > > > +    OpenEXR/IlmImf/.libs/libIlmImf.a \
> > > > +    OpenEXR/IlmImfUtil/.libs/libIlmImfUtil.a ],
> > > > +  [ Iex Half Imath IlmThread IlmImf config usr/include/OpenEXR ])
> > > >
> > > > ---
> > > >
> > > > I force-disabled openEXR probe so it will NOT pick up system
> > headers/libs,
> > > > discovered new includes demand -std=c++11 in CFLAGS :}
> > > >
> > > > seems to work ....
> > > > --
> > > > Cin mailing list
> > > > Cin at lists.cinelerra-gg.org
> > > > https://lists.cinelerra-gg.org/mailman/listinfo/cin
> > > >
> > >
> >
> >
> > --
> > Cin mailing list
> > Cin at lists.cinelerra-gg.org
> > https://lists.cinelerra-gg.org/mailman/listinfo/cin
> >
>

More information about the Cin mailing list