Sorry for the top-post, but github generates the diff from a pull request too 
they just don't publicize it:

https://github.com/OpenImageIO/oiio/pull/118.diff


On Jul 7, 2011, at 9:41 AM, Richard Shaw wrote:

> On Thu, Jul 7, 2011 at 11:22 AM, Chris Foster <[email protected]> wrote:
>> On Thu, Jul 7, 2011 at 11:33 PM, Richard Shaw <[email protected]> wrote:
>>>>> 3. There's no man pages for the binaries. It's not an absolute
>>>>> requirement for Fedora but it's *STRONGLY* recommend for any command
>>>>> line program.
>>>> 
>>>> From work that Larry just posted it looks like idiff, iinfo, iprocess,
>>>> and possibly iconvert, maketx may dissappear in the near future, so I
>>>> doubt there will be much interest in writing comprehensive docs for
>>>> these right now.  Would it be sufficient for a quick & dirty job to use
>>>> txt2man on the --help output?
>>> 
>>> That should do it. The packaging guidelines only require that they exist.
>> 
>> Ok, I spent some time (more than expected) getting this working.  The
>> build system should now autogenerate man pages on unix systems when
>> txt2man and python can be found.  (patch attached. for reviewers, see the
>> pull request as before, https://github.com/OpenImageIO/oiio/pull/118)
> 
> Can I bug you for a diff one more time? :) Would the replace the first
> one you sent me or be in addition to?
> 
> 
>>> 6. There are a number of warnings in the build but no errors. Is
>>> anyone interested in reviewing them to see if they actually need to be
>>> fixed?
>>> 
>>> What I did was to create a pseudo-index by grepping through the build
>>> log for "warning" with line numbers since anyone looking into the
>>> warning would probably need to see the lines above and/or below the
>>> warning.
>>> 
>>> I can post both of those files online somewhere if anyone's interested.
>> 
>> Feel free to post them.  I may not find time to fix them personally
>> though perhaps someone else will.
> 
> Here's the link:
> http://hobbes1069.fedorapeople.org/OpenImageIO/
> 
> oiio-build.log is the full build log
> oiio-warning.log is the "index"
> 
> 
>>> Most of these can be ignored but I'm concerned about:
>>> OpenImageIO.x86_64: W: private-shared-object-provides
>>> /usr/lib64/python2.7/site-packages/OpenImageIO.so
>>> OpenImageIO.so()(64bit)
>>> 
>>> The description of the error:
>>> $ rpmlint -I private-shared-object-provides
>>> private-shared-object-provides:
>>> A shared object soname provides is provided by a file in a path from which
>>> other packages should not directly load shared objects from.  Such shared
>>> objects should thus not be depended on and they should not result in 
>>> provides
>>> in the containing package.  Get rid of the provides if appropriate, for
>>> example by filtering it out during build.  Note that in some cases this may
>>> require disabling rpmbuild's internal dependency generator.
>> 
>> Haha!  This is the most obtuse explanation I've read in a while.  I've
>> read it a few times now, and I must say I have absolutely no idea what
>> it's telling me, and whether I should be worried.  I even looked it up
>> online, but no joy.
> 
> I almost explained it a bit further (I don't completely understand it
> though) but I figured my email was already getting rather long and
> figured it you didn't understand it you'd say so! :)
> 
> A little background: rpm is pretty good at figuring out what's needed
> to run a package, in the spec file those are the "Requires:" lines.
> You only add these if, for whatever reason, rpm couldn't figure it out
> for you. This happens most often with apps written in interpreted
> languages (python, perl, etc).
> 
> The other half of the dependency resolution equation is what a package
> "provides". Which it also tries to do automatically. This allows for
> depending on a specific binary or library instead of the package as a
> whole.
> 
> In this case it thinks that "OpenImageIO.so" is a private library
> since it's in the site-packages directory instead of one that's auto
> searched like /usr/lib{,64}. But for some reason a "Provides" is being
> generated either because rpm thinks it should be or something in the
> library is saying it does (I'm not sure which).
> 
> There are methods of filtering Provides but it's messy.
> 
> Hopefully that helps your understanding!
> 
> Thanks,
> Richard
> _______________________________________________
> Oiio-dev mailing list
> [email protected]
> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to