Commits: http://github.com/jeremyselan/OpenColorIO/commit/f5a80ee9e46eab956de7d618abda866af81de0e7 http://github.com/jeremyselan/OpenColorIO/commit/9859c936bd4c17651ec2a4ae33fa23c6476fc6d2 These

;) Thats good. It should be trying to do it all in the build/ directory, so this is a little strange. I have included all the additional libs for the html build, and install of Latex is the only

Whoa!! These are beautiful docs, thanks for the hard work! The pdf is awesome, as is the html. I think this will inspire me to keep all docstrings up to date from now on. When I run with your

I have added sphinx OCIO document generation using sphinx, there are demo's of the output here: html - http://malcolmhumphreys.github.com/OpenColorIO/ pdf -

The new 'Context' class is now exposed in python, and can be passed as an optional kwarg in

LGTM

I guess I was mainly thinking about speed, I have seen people create a hell of a lot of environment variables. But your right we can add that complexity later when / if that becomes a

Issue: http://github.com/imageworks/OpenColorIO/issues#issue/48 Commit: http://github.com/jeremyselan/OpenColorIO/commit/a0fb960d0e8b083acc4bd18048d24ac8e49f99ad All the classes that relied on the

Thanks, I've addressed all comments and rolled it into master. Fixed. Currently, nothing causes it to be flushed. Adding a library-wide "flush cache" call is at the top of my todo list. (It was

I really like most of this change, I have a few comments which are mostly cosmetic. 'Default' seems a little redundant in this names. config.getDefaultContext() vs

Commits: 1a54ec8d76 Resolves Issue #16: http://github.com/imageworks/OpenColorIO/issues#issue/16 This commit enables color transforms to leverage environment variables + search paths to determine

Thanks, I'm going to check this in then. -- Jeremy <malcolmh...@...> wrote:

yeah I guess you would either need to do that or have some config.getErrorMsg() function. No I think it's ok esp if people are used to the ilm code. The ilm code does call header.sanityCheck when

Either of our approaches will work, of course, but I'm still inclined to go with the exception route. A bunch of calls in OCIO can throw exceptions (the common one being config->getProcessor(...) ),

LGTM

Not really. Would you ever expect to call OCIO::GetCurrentConfig() and get back a pointer to an invalid profile? Throwing an exception from inside GetCurrentConfig() feels sensible, while throwing in

Malcolm - interesting idea! I do agree that right now families feel a bit like a second-class concept, and I would like them to be easier to deal with. I also agree that it's appealing to present

This checkin removed config.getEditableColorSpace(...). Users should use the existing getColorSpace, and addColorSpace as replacements. getEditableColorSpace has always been a wart on the OCIO API,

Your idea would also work. Not sure if either is better. sanityCheck() was inspired by OpenEXR, which has a function of the same name, and performs a similar action. Guess I just had that model in

Could we call this config.validate(), I know we are all a bit insane ;) enum ConfigValid { CONFIG_UNKNOWN = 0, CONFIG_VALID, CONFIG_INVALID } What do you think about not throwing an