-
Notifications
You must be signed in to change notification settings - Fork 102
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add describe-guard primitive #851
base: master
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great job @mightybyte !
In addition to comments, we just need some basic exercising of the code in tests/pact/caps.repl
-- you'll see a ton of guard stuff in there you can grab. The model would just be
(expect "describe-guard for keyset"
{ 'type: "keyset", 'keyset: (read-keyset 'k) }
(describe-guard (read-keyset 'k)))
etc
GKeySet ks -> | ||
computeGas' gas i (GUserApp Defun) $ | ||
return $ toTObject TyAny def | ||
[ ("type", tStr "KeySet") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How should we handle type-specific metadata? I'm guessing we should only have type
be common and everything else varies, but for some that's JSON no-no. What about a details
field which is a varying object? Or is that overthinking it
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The main motivating purpose for this was providing information to apps / wallets / etc about what keys they should sign with. So the big thing my immediate use case wants from this is public keys. Beyond that I really don't know. Probably the most important criteria is that any changes we make down the road can be done in a backwards-compatible manner.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess one other high level question is whether we should put the case-specific data at the same level or if we should make some kind of generic object wrapper to make it easier to decode.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
case-specific data at the same level or if we should make some kind of generic object wrapper to make it easier to decode
That was what I was trying to suggest with details
, let's do that.
src/Pact/Native/Keysets.hs
Outdated
GModule _ -> computeGas' gas i (GUserApp Defun) $ | ||
return $ toTObject TyAny def [ ("type", tStr "Module") ] | ||
GUser _ -> computeGas' gas i (GUserApp Defun) $ | ||
return $ toTObject TyAny def [ ("type", tStr "User") ] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK this is where it gets interesting.
First, it should have fun
and args
from UserGuard
.
(Note that all of these (including other types) should probably try to stay consistent with JSON field names. This is maybe why a details
field could be good, to make it really clear that we're enumerating the constructor JSON-style)
Second: wasn't it a motivation to show all keysets? This is where we might be able to traverse the guard structure. Let's look at the JSON of a gas station with a key:
{ "fun": "util.guards1.enforce-guard-any"
"args": [
[ { "pred": "keys-all",
"keys": [ "9a4849687cbcfeb1f7c6510539638da576289508aedcc75f4d6ad3ed2623635c" ]
},
{ "fun": "util.guards1.enforce-guard-all",
"args": [
[ { "fun": "coin.gas-only",
"args": []
},
{ "fun": "util.guards1.enforce-below-or-at-gas-price",
"args": [ 1e-08 ]
},
{ "fun": "util.guards1.enforce-below-or-at-gas-limit",
"args": [ { "int": 10000 } ]
}
] ]
}
] ]
}
So, first, it might be nice to try to pretty-print in a desc
field. More importantly, we can walk the args in that any time we find a TGuard
argument, we can find keysets! Then we dump them in a keysets
field. What do you think?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This sounds promising! I'm also thinking about the scenario where maybe you have a smart wallet that allows different keysets in different situations. Is there a way the smart wallet module could somehow expose what keys it's accepting? I'm imagining situations where it's not necessarily static. For instance, maybe you have a smart wallet designed to make funds available to beneficiaries on the event of death or loss of owner's keys. In this situation you might have a time delay. If it's been less than a year since the last transaction, only allow the owner's key. Otherwise allow some multi-sig of social recovery with maybe a family member plus an attorney or something.
Co-authored-by: Stuart Popejoy <[email protected]>
Co-authored-by: Stuart Popejoy <[email protected]>
@@ -49,6 +53,14 @@ keyDefs = | |||
"Define keyset as NAME with KEYSET, or if unspecified, read NAME from message payload as keyset, \ | |||
\similarly to 'read-keyset'. \ | |||
\If keyset NAME already exists, keyset will be enforced before updating to new value." | |||
,setTopLevelOnly $ defGasRNative "describe-keyset" descKeySet | |||
(funType tTyObjectAny [("keyset",tTyString)]) | |||
["(env-data {\"k\": { \"pred\": \"keys-all\", \"keys\": [\"abc\"] }) (describe-keyset (read-keyset 'k))"] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Not sure why this example is failing the test suite.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Here's working code (will need quote escapes):
(env-data {"k": ["abc"] }) (define-keyset 'k (read-keyset 'k)) (describe-keyset 'k)
@sirlensalot This feature never landed. Do we still want this? |
Also moves
describe-keyset
to the "Keysets" section, which seems more intuitive.