I think a better solution might be too use luxconsole for previews, I think blender is a moving target and once again we might be making a major change just to respond to it and for cycles of all things which we have no obligation to support, how many times have we gotten stuck with this in the past? how do we know this is the last case? How do we know what hidden problems will emerge such as the issues with network rendering? Honestly I am fairly sure this is a result of sloppy practice on the part of cycles so I don't think this is a good precedent.
Just for myself this is going to break my workflow and set me back perhaps weeks if I have to upgrade my distro on all my machines even tho I don't use pylux, which I don't mind doing if there is a Actual reason. I really think we need to think carefully about supporting stable workflows rather than create work for people without a good reason. This has nothing to do with new features either which almost never depend on boost. I can see upgrading to 1.47 if we need some of its features.
The question is will this change benefit people trying to do renders?
To me this whole situation is surreal and I am sorry but I can't let this go forward without logging an objection for the record.
Just some thoughts