Martin Atkins

The overall idea here is to move away from storing the whole state as a single blob and acquiring an exclusive lock on the entire state at once, and instead...

Although I've not investigated this in detail _at all_ yet, an interesting consequence of tracking a separate blob for each object rather than a single blob for the entire state...

While discussing this today I realised I introduced a bug in some of my most recent work: Originally I had the state objects representing whole resources (rather than resource instances)...

The new behavior was documented in https://github.com/opentofu/opentofu/pull/3063, giving the current documentation text: > The logical operators in OpenTofu are short-circuiting, meaning `var.foo == null || var.foo.bar == 1` will not...

Thanks for working on it, @thc1006! I intended my previous comment as a question about whether the other participants thought that just updating the changelog would be sufficient, so I'm...

Hi @waddles, Unfortunately I think it unlikely that the `hashcorp/aws` provider would include a write-only version of `user_data`, because the design of that provider tends to echo the design of...

This could potentially help with https://github.com/opentofu/opentofu/issues/1784, although I'm not 100% sure since that feature request is not very explicit about exactly what intermediate data would've been helpful in debugging the...

As I noted in the writeup, I'm kinda skeptical about interactive debugging in the middle of a `tofu plan` just because the order of execution is not necessarily deterministic and...

The connection with `tofu console` also made me realize that in https://github.com/opentofu/opentofu/pull/2304 I introduced something called `addrs.ExprScope` that is a general way to represent "place where expressions can be evaluated"...

It seems like we have a number of different ideas to explore related to exposing intermediate values for debugging, and that we might want to do something a little different...