Fix loading of CreateViews nodes saved from Gaffer 1.5+ #6059
Conversation
- Fix error when resizing removed a plug with an input connection. In this case, the "resize when inputs change" behaviour was kicking in during the outer resize and messing everything up. - Fix failure to resize output plugs, due to attempting to add an input plug as a child.
In 1.5, the `views` array is created with a prototype, and `resize()` is used in the serialisation to add all necessary children. When running `resize()` in 1.3 though, we have no prototype and can't resize. So we make `ArrayPlug::resize()` a no-op in this case and add the plugs on a just-in-time basis in a compatibility config. I really don't like monkey-patching all ArrayPlugs like this, but it's not possible to monkey-patch individual instances when patching Python's magic methods (of which `__getitem__` is one). We'll rip out all of this code when merging forwards to `main` though, so it won't live for _too_ long.
|
Oh, that is kind of unfortunate ... sorry for causing this mess when I wrote CreateViews :( Looks like a reasonable solution though, and it's good that it won't be needed on main. LGTM. Test failures might require a bit of attention though ... does our whitespace checker not like how we write out .gfr files? And have we ever seen it before that |
Yeah, there's a trailing newline it doesn't like. Completely harmless, so I haven't yet found the motivation to deal with it.
Yeah, I've seen this a couple of times now - nothing to do with this PR. I will need to find some time to bang my head against a Windows build for a bit sometime soon...but I'm already pretty sure the problem is in the test and not the thing being tested. |
We can share responsibility here - I reviewed it, and I was the one who put the loopholes in ArrayPlug in the first place. |
Back in #6015, we improved the way we created and serialised ArrayPlugs. For the most part this all works fine with earlier Gaffer versions, but it created serialisation for CreateViews that would crash in Gaffer 1.3 and 1.4. This PR backports some of the bugfixes from that PR, and then adds a compatibility shim to allow the files to load. As noted in the commit log, I'm not especially happy with this approach, but it will at least be short-lived.