bicepBCP422 - The warning is unclear

[slavizh]

Bicep version
Bicep CLI version 0.36.177 (09988bb)

Describe the bug
I cannot understand the intent of this warning:

A resource of type "Microsoft.Storage/storageAccounts | null" may or may not exist when this function is called, which could cause the deployment to fail.bicep[BCP422](https://aka.ms/bicep/core-diagnostics#BCP422)

First it is unclear why this warning appears at all. Especially that I have condition on calling the function so I know if I should call the function when the account exists and when it does not. Also it is unclear what I need to do to avoid this warning. If the only option is to suppress this manually what is the point of the rule at all?
To Reproduce
Steps to reproduce the behavior:

Additional context
Add any other context about the problem here.

[slavizh]

On top of this the warning appears on module outputs and properties of resources as well. It is very annoying.

[slavizh]

@jeskew any news as it is pretty annoying warning and appears almost in every solution. Additionally I would like to propose to have a way to disable these with bicepconfig.json. I have checked the existing ones https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/linter and I couldn't find rule for it. I think that for every rule we should be able to disable it at scale if needed.

[jeskew]

The BCP422 warning does have a quickfix that will use a .? operator if that helps. The last Bicep release (which introduced this check, meant to address #3750) also updated how the .? operator works on resource property access -- it will now use the resource's condition as a guard.

Cases like the storageContainerPath property in the screenshot above need something more sophisticated since Bicep doesn't have any equivalent of lexically scoped pattern matching assignment like in C#. One option here would be a flag in bicepconfig to turn off nullability analysis

[slavizh]

@jeskew if I already have a condition (which I have) it should be able to detect that and not produce warning. After all I may choose different approach than using .?. To be honest I think these warnings are getting out of hand. Either they should work flawlessly or there should be an option to disable them. It gets really weird for end users who deploy templates without having knowledge on bicep. What adds to this is that in pipeline we see these 2 times - once we compile the bicep parameters files and another time when we do the deployment. There is no way to show only warnings related to bicep parameters file and we want to compile it before deployment so end user can see the actual configuration deployed due to usage of variables and other functions in the bicepparam file.

[jeskew]

@slavizh You can suppress any nullability warning with a !, but that will only apply to a single value. I'll write up a proposal for how to disable nullability analysis like you can do in C# and present to the team.

[slavizh]

@jeskew that I understand but requires updating a hundreds of templates :)