Add a concept of a caveat TypeSet to allow overriding the types available to caveat processing

[josephschorr]

No description provided.

[barakmich]

I like the fact it's all piped through! But then it immediately reads to me as though there's a config struct waiting to happen for the params to some of the NewX functions, specifically in internal/dispatch/graph/graph.go and internal/graph/lookupresources2.go (and maybe internal/services/v1/schema.go)

[josephschorr]

I like the fact it's all piped through! But then it immediately reads to me as though there's a config struct waiting to happen for the params to some of the NewX functions, specifically in internal/dispatch/graph/graph.go and internal/graph/lookupresources2.go (and maybe internal/services/v1/schema.go)

Almost certainly. Do you think we should do it now or as a followup?

[barakmich]

I like the fact it's all piped through! But then it immediately reads to me as though there's a config struct waiting to happen for the params to some of the NewX functions, specifically in internal/dispatch/graph/graph.go and internal/graph/lookupresources2.go (and maybe internal/services/v1/schema.go)

Almost certainly. Do you think we should do it now or as a followup?

Might as well; easy enough to find the locations since it follows this diff

[tstirrat15]

LGTM, pending removal of the zip and a question

[tstirrat15]

Unintentional?

[josephschorr]

Yep

[tstirrat15]

Is this a convention? Where does this actually prevent the TypeSet from being modified?

[josephschorr]

Check below: it can't be used if not frozen

[tstirrat15]

Right, it can't be used, but nothing prevents me from touching a frozen typeset (unless I'm missing something)

[tstirrat15]

LGTM

[barakmich]

Boy howdy that's a lot of files. Fortunately, mostly signature-driven.

LGTM