ok, Ive found out whats going on with the first issue, though not quite the cause...
and Ive actually seen this a few times now, not just on this patch
@johannes this is fixed by re-saving the sub-patch (warning: don't do this for factory object guys!)
we can then diff them and we see:
- <obj type="mix/mix 1 g" sha="9837ebd6f7c0b2b3853ea475d91c943144e2273b" name="sub_osc_mix" x="672" y="70">
+ <obj type="mix/mix 1 g" uuid="e6982841c1bf323ee2062a4b6cc2737adafbd668" name="sub_osc_mix" x="672" y="70">
ok, this is not totally unexpected... so Im not surprised about the fact sha has gone and uuid is there. but what I am surprised about is that the original did not have the UUID...
now obviously, the code is kind of doing the correct thing, its looking up using the name, and if we open in the editor all is well (and hence save is working) BUT it would appear its not totally correct....
my first thought is perhaps its picking up the incorrect version of mix 1g and for some reason, opening in the editor is re-evaluating which object to use, perhaps using type matching?
It should be easy enough to debug, we just need to see when a subpatch is loaded for going live, its not evaluating in the same way as when editing the subpatch.
but slightly concerned there may be deeper issues....
Id say this is show-stopper for release, as I'm assuming it will probably affect many patches,
actually Im very surprised we haven't had more testers reporting it, Id say its a pretty common issue.
unfortunately, I don't have time to look at it today, as I'm out all day... and not sure about tomorrow, as things still pretty hectic.
you other issue @keyman ... yeah there are a few things that need to be locked down when live... I guess we need to log it and fix. the case of the UI issue, is it not being able to find the inlet/outlet, the question is why