OK, I've heard no serious objections, a few "either 1.1 or 1.2 is ok", and
several "just do it now."
So I'm throwing the switch.
I apologize again for any confusion resulting from changing the API after
tagging the release. Hopefully, only one week after release, the damage is
limited. We'll try not to do it again.
-- lg
On Nov 16, 2012, at 11:15 AM, Larry Gritz wrote:
> There's a proposal on the table to change the way a few ImageInput functions
> (the read varieties that have channel subsets) interpret those channel
> parameters: switching from (first,n) to (begin,end). We *think* that these
> particular functions are rarely called (they were only added in 1.0 and not
> necessary for most use), and when they are they almost always fall into two
> use cases -- all channels, and one channel -- both of which will behave
> identically with the new semantics.
>
> In particular, I'm tempted to quickly add this to 1.1, despite generally
> wanting to avoid API changes of a release branch -- but since it was just
> released last week, I think we still have the opportunity without disrupting
> too many people. I could save it until 1.2, but that just gives people more
> opportunity to become dependent on that rare case whose meaning will change.
>
> Anyway, several smart people are independently telling me to go ahead and
> pull the trigger and add it to 1.1, but I wanted to give everybody else one
> last chance to talk me out of it.
>
> If you care, here's the pull request to read the discussion and comment:
> https://github.com/OpenImageIO/oiio/pull/466
>
--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
