WIP: Draft version constraint generation #5472

Gertjan423 · 2020-04-07T16:42:02Z

First draft version of static verification tool.

Reads DAR files.
Partially evaluates the code.
Generates constraints for all fields, as the tool does not yet read annotations regarding which fields should be tracked.

Main TODOs at the moment:

Implement some undefined's and TODO's throughout the code.
Take additional inputs or read annotations to determine which fields should be tracked.
Extend with additional language constructs (recursion).
Additional testing.

cocreature

Awesome, congrats on your first PR @Gertjan423 🎉 A few preliminary comments but I very much like the direction this is going in!

cocreature · 2020-04-07T16:43:48Z

compiler/daml-lf-verify/daml-lf-verify.cabal

@@ -0,0 +1,57 @@
+cabal-version: 2.4


You don’t need a cabal file, we build our code with Bazel.

cocreature · 2020-04-07T16:44:11Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify.hs

+
+-- | Static verification of DAML packages.
+module DA.Daml.LF.Verify
+  ( module LF


Why do you reexport LF if this is main?

cocreature · 2020-04-07T20:04:34Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+genForExercise tem ch cid par arg = do
+  cidOut <- genExpr cid
+  argOut <- genExpr arg
+  -- TODO: Take possibility into account that the choice is not found?


That would be caught by the LF typechecker, so no need to do more than crash here.

cocreature · 2020-04-07T20:06:47Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+    -- TODO: This is a random error, as we do not have access to the expected
+    -- type, which we need to constructed the error we really want.
+    -- Perhaps we do need to define our own set of errors.
+    -- _                      -> throwError (EExpectedRecordType ty)


We generally format to be somewhat friendly to diffs. So no alignment, line breaks and indentation such that the amount of indentation does not depend on identifier length. Not a hard rule but I like being somewhat consistent here.

cocreature · 2020-04-07T20:11:30Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+  ETyApp expr typ -> genForTyApp expr typ
+  EVar name       -> genForVar name
+  EVal w          -> genForVal w
+  EUpdate (UCreate tem arg)              -> genForCreate tem arg


This doesn’t seem quite right to me. You are conflating evaluation and execution afaict. Update works like IO in Haskell. Evaluating an Update expression does not do anything. The way this is implemented at runtime is by passing around an opaque token (similar to the RealWorld token in Haskell) so Update expressions are actually functions. You can probably do something similar for your analysis: If you start from a choice you look at the update expression evaluated to the token. If you just see an Update expression that is not applied to the token, you don’t look inside.

cocreature · 2020-04-07T20:21:38Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+--   + or create our own world environment to store values. This also makes
+--   sense as these values work similar to how they work in the type checker,
+--   except that we need to store a partially evaluated definition as well.
+-- Both approaches make sense, but both imply a lot of code duplication, so they


I wouldn’t worry too much about duplicating code for now.

cocreature · 2020-04-07T20:25:58Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+    -- abstract over it. Same for this returning foldM.
+    genPackage' :: MonadDelta m => Delta
+                -> (PackageId, (Package, Maybe PackageName)) -> m Delta
+    genPackage' deli (idi, (paci, _)) =


Why do all variable names end with i? 🙂

cocreature · 2020-04-07T20:29:29Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+                -> (PackageId, (Package, Maybe PackageName)) -> m Delta
+    genPackage' deli (idi, (paci, _)) =
+      -- TODO: This is getting quite unreadable.
+      foldM (\deli' modi -> genModule' deli' (PRImport idi, modi))


I’m probably missing something but it looks to me like this only works if modules are sorted topologically. Otherwise, you might not be able to resolve a cross-module reference since you haven’t yet included the delta for that module?

cocreature · 2020-04-07T20:31:35Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+  let qname = Qualified pac mod (fst $ dvalBinder val)
+  return $ set devals (singleton qname (expOut ^. goExp, expOut ^. goUpd)) emptyDelta
+
+-- TODO: Handle annotated choices, by returning a set of annotations.


What is an annotated choice? Presumably that’s some new annotation that you want to add to express constraints that should be verified?

I’m still not quite sure what an annotated choice is 🙂

cocreature · 2020-04-07T20:35:21Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+                                      (set usCre [UpdCreate tem fs] emptyUpdateSet)
+                                      (argOut ^. goDel))
+    _                      -> throwError EEnumTypeWithParams
+    -- TODO: This is a random error, as we do not have access to the expected


I’m perfectly happy with just calling error for now for anything that isn’t well-typed LF.

I usually use something ala error "IMPOSSIBLE: Ill-typed DAML-LF" in situations like this one.

The impossible happened 😮

hurryabit · 2020-04-08T07:56:54Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+    -- ^ The skolemised term variables.
+  , _devals :: !(HashMap (Qualified ExprValName) (Expr, UpdateSet))
+    -- ^ The bound values.
+  , _dchs :: !(HashMap ((Qualified TypeConName), ChoiceName) UpdateSet)


There are a few more unnecessary parentheses around, but hlint should catch that in the end anyway.

Suggested change

, _dchs :: !(HashMap ((Qualified TypeConName), ChoiceName) UpdateSet)

, _dchs :: !(HashMap (Qualified TypeConName, ChoiceName) UpdateSet)

hurryabit · 2020-04-08T08:02:43Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+lookupDExprVar x = ask >>= \ del -> unless (elem x $ del ^. devars)
+                                          $ throwError $ EUnknownExprVar x


I don't think lenses buy us anything here. I also think a >>= \x -> ... is a sign to use do notation.

Suggested change

lookupDExprVar x = ask >>= \ del -> unless (elem x $ del ^. devars)

$ throwError $ EUnknownExprVar x

lookupDExprVar x = do

del <- ask

unless (x `elem` _devars del) $ throwError $ EUnknownExprVar x

hurryabit · 2020-04-08T08:06:06Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+  = extendGOUpds (genOut2 ^. goUpd)
+  $ extendGODelta (genOut2 ^. goDel)
+    genOut1


Isn't that just plain function application?

Suggested change

= extendGOUpds (genOut2 ^. goUpd)

$ extendGODelta (genOut2 ^. goDel)

genOut1

= extendGOUpds (_goUpd genOut2)

$ extendGODelta (_goDel genOut2)

genOut1

hurryabit · 2020-04-08T08:07:35Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+      -- TODO: This is getting quite unreadable.
+      foldM (\deli' modi -> genModule' deli' (PRImport idi, modi))
+            deli (NM.toList $ packageModules paci)
+      >>= return . (concatDelta deli)


act >>= return . f is the same as fmap f act or f <$> act, depending on what you prefer.

hurryabit · 2020-04-08T08:08:59Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+  ETmApp fun arg  -> genForTmApp fun arg
+  ETyApp expr typ -> genForTyApp expr typ
+  EVar name       -> genForVar name
+  EVal w          -> genForVal w


Please don't align the arrows in case expressions. This will lead to a lot of "diff noise" down the line.

hurryabit · 2020-04-08T08:10:44Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+genValue pac mod val = do
+  expOut <- genExpr (dvalBody val)
+  let qname = Qualified pac mod (fst $ dvalBinder val)
+  return $ set devals (singleton qname (expOut ^. goExp, expOut ^. goUpd)) emptyDelta


Suggested change

return $ set devals (singleton qname (expOut ^. goExp, expOut ^. goUpd)) emptyDelta

return $ emptyDelta{_devals = singleton qname (_goExp expOut, _goUpd expOut)}

hurryabit · 2020-04-09T07:42:01Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify.hs

+  Options{..} <- execParser optionsParserInfo
+  pkgs <- readPackages optInputDars
+  let _delta = runDelta $ genPackages pkgs
+  putStrLn "Constraints generated."


We need to deeply force _delta here, otherwise nothing will really happen in the test runs. Maybe just print it?

hurryabit · 2020-04-09T07:44:36Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+concatDelta (Delta vars1 vals1 chs1) (Delta vars2 vals2 chs2) =
+  Delta (vars1 ++ vars2) (vals1 `union` vals2) (chs1 `union` chs2)
+  -- TODO: union makes me slightly nervous, as it allows overlapping keys
+  -- (and just uses the first)


I guess you want unionWith concatUpdateSet.

hurryabit · 2020-04-09T07:45:42Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+introDelta delta = local (concatDelta delta)
+
+-- TODO: This is a bit strange in a reader monad.
+-- Figure out a way to extend, instead of overwrite every time.


Maybe withDelta would communicate your intent better?

hurryabit · 2020-04-09T07:47:24Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+setDelta delta = local (const delta)
+
+extVarDelta :: MonadDelta m => ExprVarName -> m a -> m a
+extVarDelta x = local (over devars ((:) x))


Would it make sense to have a function

varDelta :: ExprVarName -> Delta

and use introDelta (varDelta x) instead of extVarDelta x? At least to me, these primitives feel "more orthogonal" to me.

hurryabit · 2020-04-09T07:54:35Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+-- TODO: Random error.
+
+-- | Helper functions mirrored from Env.
+-- TODO: Reduce duplication by abstracting over MonadGamma and MonadDelta?


Nah, I wouldn't bother with it for now. You don't want to tie yourself into unnecessary constraints. 😃

hurryabit · 2020-04-09T08:20:17Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+  = extendGOUpds (_goUpd genOut2)
+  $ extendGODelta (_goDel genOut2)
+    genOut1


If you write this in a more direct style, you can remove a fair bit of code below:

Suggested change

= extendGOUpds (_goUpd genOut2)

$ extendGODelta (_goDel genOut2)

genOut1

= GenOutput

{ _goExp = _goExp genOut1

, _goDel = _goDel genOut1 `concatUpdateSet` _goDel genOut2

, _goUpd = _goUpd genOut1 `concatUpdateSet` _goUpd genOut2

}

hurryabit · 2020-04-09T08:22:33Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+    -- ^ The updates, performed by this expression.
+  , _goDel :: Delta
+    -- ^ The context extension, made by this expression.
+  }


In my experience, spelling out identifiers has proven beneficial most of the time, ie., _goExpr, _goUpdate and _goDelta. (Expr is such a common unambiguous abbreviation that it's ok, whereas Del could also abbreviate "delete".)

hurryabit · 2020-04-09T09:09:01Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+buildDelta :: MonadDelta m => Delta -> (a -> m Delta) -> [a] -> m Delta
+buildDelta del op args = foldM step del args
+  where
+    step d x = concatDelta d <$> setDelta d (op x)


My suspicion, fuelled by your comment on setDelta, is that you want something like this:

introDeltas :: MonadDelta m => (a -> m Delta) -> [a] -> m b -> m b introDeltas _ [] act = act introDeltas f (x:xs) act = do delta <- f x introDelta delta (introDeltas f xs act)

and then have functions like

introModule :: MonadDelta m => PackageRef -> Module -> m b -> m b

Is that correct?

hurryabit · 2020-04-09T09:13:18Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+                                      (set usCre [UpdCreate tem fs] emptyUpdateSet)
+                                      (argOut ^. goDel))
+    _                      -> throwError EEnumTypeWithParams
+    -- TODO: This is a random error, as we do not have access to the expected


I usually use something ala error "IMPOSSIBLE: Ill-typed DAML-LF" in situations like this one.

hurryabit · 2020-04-09T09:14:34Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+    -- ^ The list of archive updates.
+  , _usCho :: ![UpdChoice]
+    -- ^ The list of choice updates.
+  }


Don't we need to account for fetches as well?

cocreature · 2020-04-09T09:31:47Z

@Gertjan423 CI is currently failing due to the format check. You can easily fix that by running ./fmt.sh locally or if you want to have it be a bit faster ./fmt.sh --diff. The latter is fast enough that some people even set it up for a precommit or at least prepush hook.

changelog_begin changelog_end

cocreature · 2020-05-07T19:26:19Z

WORKSPACE

@@ -784,6 +784,15 @@ nixpkgs_package(
    repositories = dev_env_nix_repos,
 )

+nixpkgs_package(


Note that once we want to actually make this a feature we ship to users we do need to support Windows. But let’s worry about this after we get this merged.

cocreature · 2020-05-07T19:28:34Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Read.hs

+
+data Options = Options
+    { optInputDar :: FilePath
+    , optChoiceTmpl :: String


Might be easier to understand if you already use the proper LF type here.

cocreature · 2020-05-07T19:29:20Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify.hs

+      choiceName = ChoiceName (pack optChoiceName)
+      fieldTmpl = TypeConName [pack optFieldTmpl]
+      fieldName = FieldName (pack optFieldName)
+  solver <- locateRunfiles "z3_nix/bin/z3"


Note that this only works in our Bazel setup. We need to figure out what we do if we ship things to users. Not sure if we want to ship an SMT solver so a reasonable option might be to let users pass a path to the SMT solver they want to use.

compiler/daml-lf-verify/src/DA/Daml/LF/Verify.hs

cocreature · 2020-05-07T19:30:53Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Read.hs

+readPackages :: [FilePath] -> IO [(PackageId, (Package, Maybe PackageName))]
+readPackages dars = concatMapM darToPackages dars
+  where
+    darToPackages :: FilePath -> IO [(PackageId, (Package, Maybe PackageName))]


We seem to be repeating that piece of code over and over. We should probably move it into some helper library at some point (not in this PR).

cocreature · 2020-05-11T08:49:41Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

+  -- ^ Function for looking up references in the update set.
+  -> (updset -> ([Cond ref], updset))
+  -- ^ Function popping the references from the update set.
+  -> (updset -> updset -> updset)


Can we give this a Semigroup instance?

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Context.hs

cocreature · 2020-05-11T08:52:10Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Generate.hs

+
+-- | Analyse a case expression.
+-- TODO: Atm only boolean cases are supported
+genForCase :: (GenPhase ph, MonadEnv m ph)


SMT does have ADTs so we could in principle use those but you can also encode pattern matches without using them.

cocreature · 2020-05-11T08:55:17Z

compiler/daml-lf-verify/src/DA/Daml/LF/Verify/Solve.hs

+  log <- S.newLogger 1
+  sol <- S.newSolver spath ["-in"] (Just log)
+  vars <- declareVars sol $ filterVars _cVars (_cCres ++ _cArcs)
+  cre <- foldl S.add (S.real 0) <$> mapM (cexp2sexp vars) _cCres


How does this work? You are working with integers in some places.

I think it works somewhat by accident because 0 is both a real and an integer literal. We should check the type.

cocreature · 2020-05-11T09:00:23Z

compiler/daml-lf-verify/tests/DA/Daml/LF/Verify/Tests.hs

+      verify condDar False tmpl choice tmpl field >>= \case
+        Success -> assertFailure "Verification wrongfully passed for FailA - content"
+        Unknown -> assertFailure "Verification inconclusive for FailA - content"
+        Fail _ -> return ()


Ar the results not deterministic or why are you not just using assertEqual on the full Fail expression?

Co-authored-by: Moritz Kiefer <moritz.kiefer@purelyfunctional.org>

hurryabit

There are a few files that should be deleted.

There's also the change to Numeric. I'd like to understand why we do this before we merge. If you can undo this change and everything still works, please do so and feel free to merge.

compiler/daml-lf-verify/daml/conditionals/.dlint.yaml

compiler/daml-lf-verify/daml/conditionals/daml/Main.daml

compiler/daml-lf-verify/daml/conditionals/daml/Setup.daml

compiler/daml-lf-verify/daml/conditionals/frontend-config.js

compiler/daml-lf-verify/daml/conditionals/pom.xml

compiler/daml-lf-verify/daml/conditionals/.gitignore

compiler/daml-lf-verify/daml/conditionals/daml.yaml

compiler/daml-lf-verify/daml/quickstart/.gitignore

...er/daml-lf-verify/daml/quickstart/src/main/java/com/digitalasset/quickstart/iou/IouMain.java

compiler/daml-lf-verify/daml/quickstart/src/main/resources/logback.xml

hurryabit · 2020-05-19T19:11:57Z

@Gertjan423 You need to run ./fmt.sh in the root of the repo again.

compiler/daml-lf-verify/BUILD.bazel

CI currently errors with: ``` Subprocess: git checkout efe6545 -- docs/source/support/release-notes.rst failed with exit code 127; output: --- --- err: --- Previous HEAD position was 2af134c... WIP: Draft version constraint generation (#5472) HEAD is now at efe6545... 1.2.0-snapshot.20200520.4224.0.2af134ca (#6040) /bin/sh: 2: --: not found --- ``` because the line ``` latest_release_notes_sha <- shell "git log -n1 --format=%H HEAD -- LATEST" ``` will assign a string that ends in a newline, and then when we try to construct the shell command: ``` (shell_ $ "git checkout " <> latest_sha <> " -- docs/source/support/release-notes.rst") ``` we actually get two lines for Bash to execute. CHANGELOG_BEGIN CHANGELOG_END

Draft version constraint generation

040ef56

Gertjan423 requested a review from hurryabit April 7, 2020 16:42

Gertjan423 requested a review from cocreature as a code owner April 7, 2020 16:42

cocreature reviewed Apr 7, 2020

View reviewed changes

hurryabit reviewed Apr 8, 2020

View reviewed changes

Some code cleanup

0205507

hurryabit reviewed Apr 9, 2020

View reviewed changes

hurryabit marked this pull request as draft April 9, 2020 09:17

Gertjan423 and others added 21 commits April 9, 2020 14:26

Substitution functions

3476a5e

Template generation

4675fdd

Reformatted

afc3766

Attempt to fix failing check

c329b37

Rework generation for value definitions into 2 phase approach

f3e4d17

Various small changes

a7ef445

Rework delta to env state

c99827d

Incorporate data type definitions

28eb1bb

Various constraint generation bug fixes and improvements

80bb27a

First part of constraint solving

220d21d

Minor bugfixes

2bad89e

Added simple-smt to bazel

733f929

Depend on z3

5545943

changelog_begin changelog_end

Working pipeline

b736d85

Fix overwriting value definitions, and delayed choice bindings

4bc25cb

Keep track of contract id's after fetching

310d7b0

Renaming of choice parameters

4b08356

Cleanup

e04944f

Reals instead of ints

18b0b97

Various TODOs

2f3bf70

Added function documentation

0911a90

cocreature reviewed May 11, 2020

View reviewed changes

Gertjan423 and others added 11 commits May 12, 2020 16:54

Testing: build daml instead of storing dar files

64370c2

Proper handling of update types; constraint synonyms

f3bee74

Additional equality constraints after create update

13eb8a4

Make assertion print conditional

16aa3e3

Various improvements from PR feedback

642c1b9

Apply suggestions from code review

c791f31

Co-authored-by: Moritz Kiefer <moritz.kiefer@purelyfunctional.org>

Small fix: Add missing import

ba1fad5

Small fix: convert numerics

344eca8

Replace PRSelf with correct package reference

1cdf6e4

Style fixes

cdbed09

Rework conditionals to remove When

c62c276

hurryabit approved these changes May 19, 2020

View reviewed changes

Cleanup test daml files

ed91ada

Gertjan423 marked this pull request as ready for review May 19, 2020 16:26

Gertjan423 requested a review from garyverhaegen-da as a code owner May 19, 2020 16:26

Gertjan423 and others added 5 commits May 19, 2020 18:30

Merge branch 'master' into daml-lf-verification

b803eb9

Small style fixes

a3ee542

Bugfix after merge

91c989c

Attempt to fix bazel windows error

337b0ca

Attempt #2 to fix bazel windows error

49563b0

Attempt #3 to fix bazel windows error

ae4a276

cocreature reviewed May 19, 2020

View reviewed changes

compiler/daml-lf-verify/BUILD.bazel Outdated Show resolved Hide resolved

Gertjan423 added 2 commits May 19, 2020 22:03

Attempt #4 to fix bazel windows error

77ea060

Attempt #5 to fix bazel windows error

2d7c840

Gertjan423 merged commit 2af134c into master May 20, 2020

Gertjan423 deleted the daml-lf-verification branch May 20, 2020 06:08

	, _dchs :: !(HashMap ((Qualified TypeConName), ChoiceName) UpdateSet)
	, _dchs :: !(HashMap (Qualified TypeConName, ChoiceName) UpdateSet)

		lookupDExprVar x = ask >>= \ del -> unless (elem x $ del ^. devars)
		$ throwError $ EUnknownExprVar x

-lookupDExprVar x = ask >>= \ del -> unless (elem x $ del ^. devars)
-                                          $ throwError $ EUnknownExprVar x
+lookupDExprVar x = do
+  del <- ask
+  unless (x `elem` _devars del) $ throwError $ EUnknownExprVar x

	return $ set devals (singleton qname (expOut ^. goExp, expOut ^. goUpd)) emptyDelta
	return $ emptyDelta{_devals = singleton qname (_goExp expOut, _goUpd expOut)}

-  = extendGOUpds (_goUpd genOut2)
-  $ extendGODelta (_goDel genOut2)
-    genOut1
+  = GenOutput
+      { _goExp = _goExp genOut1
+      , _goDel = _goDel genOut1 `concatUpdateSet` _goDel genOut2
+      , _goUpd = _goUpd genOut1 `concatUpdateSet` _goUpd genOut2
+      }

WIP: Draft version constraint generation #5472

WIP: Draft version constraint generation #5472

Conversation

Gertjan423 commented Apr 7, 2020 • edited Loading

cocreature left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cocreature commented Apr 9, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hurryabit left a comment

Choose a reason for hiding this comment

hurryabit commented May 19, 2020

Gertjan423 commented Apr 7, 2020 •

edited

Loading