Initial CLI update for Allegra and Mary eras #2129

dcoutts · 2020-11-26T03:23:25Z

Re-implement the tx build-raw command in terms of the new era-generic TxBodyContent API.

This should allow it to work for all eras, including Byron. It checks that each feature can be used in the selected era.

There are still a number of TODOs, but this establishes the pattern.

dcoutts · 2020-11-26T03:24:04Z

This builds on #2121 and should be rebased once that PR is merged.

kevinhammond · 2020-11-26T13:58:25Z

Quick comment - should we be exposing the word "era" in the CLI? We've backed away from this in the product/marketing areas

It's more convenient this way since it means we can use it when we have a CardanoEra value or when we have an IsCardanoEra class constraint.

This lets us dynamically compare era tokens and establish if they're the same. This will be useful in the CLI for example where we read two files and need to check they're from the same era.

This pairs up some era-dependent type with a 'CardanoEra' value that tells us what era it is, but hides the era type. This is useful when the era is not statically known, for example when deserialising from a file. We will use this in the CLI when reading files, like tx, txbody and witness files. Also add InAnyShelleyBasedEra following the same pattern, which we need for Shelley-based era features like scripts.

Adjust CardanoEraStyle, InAnyCardanoEra and InAnyShelleyBasedEra to carry a corresponding class context as well as the explicit value representation of the era. This turns out to be rather useful since it lets us use functions that expect an explicit arg, or ones that expect an implicit one via a class constraint.

They tell us if the feature is available in a given era, and if so provide us with the evidence. This will make working with these features (e.g. in the CLI) easier to do in a era-general kind of way.

GADTs can only provide positive evidence, and we sometimes need positive evidence of things that are negations of each other. For example: whether the era supports multi-asset or ada-only are opposites of each other. We need to provide this positive evidence in either direction when making a TxOutValue for example data TxOutValue era where TxOutAdaOnly :: OnlyAdaSupportedInEra era -> ... TxOutValue :: MultiAssetSupportedInEra era -> ... So it is more helpful if our feature functions give us the evidence in both directions for these cases, so specifically: multiAssetSupportedInEra :: CardanoEra era -> Either (OnlyAdaSupportedInEra era) (MultiAssetSupportedInEra era) The same holds for tx fees: implicit vs explicit. Adjust the TxFee representation so it follows the same pattern with positive and negative evidence with a new type TxFeesImplicitInEra, and provide txFeesExplicitInEra :: CardanoEra era -> Either (TxFeesImplicitInEra era) (TxFeesExplicitInEra era)

Defaults to --shelley-era When we make a transaction we need to know which era it is for.

For going from an untyped UseCardanoEra enumeration to the strongly typed CardanoEra era.

We now use a per-field validation that checks that the feature is available to use in that era.

Switch the type that the command line parsers use from TxOut ShelleyEra to a new universal type TxOutAnyEra. Add CLI parser support for tx lower bounds, aux scripts and value minting. Make the tx upper bound and fee optional. Update runTxBuildRaw to use all of these, resolving several TODOs. Add validation that multi-asset tx-outs can only be used in eras supporting them, with appropriate failure messages otherwise.

For Tx, TxBody and Witness. We need to be able to read these files at the types corresponding to any of the supported eras.

Easy because all the functions they call have already been generalised.

This is slightly awkward. In principle the right thing to do is probably to do what we did with addresses and split out the language vs the era, so that we can talk about script languages independently from the eras in which those languages are supported. This would let us read scripts into some universal type and later check that the script language is supported in the era in which it is intended to be used. For now however in this PR we will hack it for now. We'll do the language vs era separation in a later PR. To do the temporary hack we add a function coerceSimpleScriptEra that lets us use the type MultiSigScript ShelleyEra as our "universal" type and the coerce function is to convert into the era. In this case we're using the minimal rather than maximal features, so the conversion is total. In this first step we're doing the identity era coersion. That will change in the next step.

These all work by reading a tx body. We use the era tx body is for to determine the era to use for everything else in each command. For a simple command like txid there's nothing else, so it will work for any era in a straightforward way. For most other commands the code can currently only support Shelley-based eras, so we add a dynamic test to fail for the Byron era. For witness and sign commands the tx body era simply determines the witness era. For assemble things are more interesting since we have to combine a txbody with a number of witnesses. We use a testEquality to check if the era of each witness matches the era of the txbody.

dcoutts · 2020-11-26T22:08:50Z

Quick comment - should we be exposing the word "era" in the CLI? We've backed away from this in the product/marketing areas

Yes. The CLI exposes the technical constraints. The tx build command needs to know the era in which the tx is intended to be used, because different features are available in different eras, and they have potentially different binary formats. This is a technicality that we cannot hide.

Needs a slight tweak to submitTx in the API to cover the Allegra and Mary era cases.

intricate · 2020-11-26T23:32:42Z

cardano-cli/src/Cardano/CLI/Shelley/Parsers.hs

+readValue :: String -> Either String Value
+readValue _maCliSyntax = Left "Need 2072 for MA cli syntax parser"


Right, I'll patch this up once we merge this PR.

intricate · 2020-11-26T23:33:58Z

cardano-cli/src/Cardano/CLI/Shelley/Parsers.hs

+pTxLowerBound :: Parser SlotNo
+pTxLowerBound =
  SlotNo <$>
    Opt.option Opt.auto
-      (  Opt.long "ttl"
+      (  Opt.long "lower-bound"
      <> Opt.metavar "SLOT"
-      <> Opt.help "Time to live (in slots)."
+      <> Opt.help "Time that transaction is valid from (in slots)."
      )

+pTxUpperBound :: Parser SlotNo
+pTxUpperBound =
+  SlotNo <$>
+    Opt.option Opt.auto
+      (  Opt.long "upper-bound"
+      <> Opt.metavar "SLOT"
+      <> Opt.help "Time that transaction is valid until (in slots)."
+      )


Maybe we should also note whether these endpoints (lower and upper bound) are inclusive or exclusive.

We should yes. It's inclusive on the lower bound and exclusive on the upper.

intricate · 2020-11-26T23:34:52Z

cardano-cli/src/Cardano/CLI/Shelley/Run/Transaction.hs

+      txBodyContent <-
+        TxBodyContent
+          <$> validateTxIns  era txins
+          <*> validateTxOuts era txouts
+          <*> validateTxFee  era mFee
+          <*> ((,) <$> validateTxValidityLowerBound era mLowerBound
+                   <*> validateTxValidityUpperBound era mUpperBound)
+          <*> validateTxMetadataInEra  era metadataSchema metadataFiles
+          <*> validateTxAuxScripts     era scriptFiles
+          <*> validateTxWithdrawals    era withdrawals
+          <*> validateTxCertificates   era certFiles
+          <*> validateTxUpdateProposal era mUpdatePropFile
+          <*> validateTxMintValue      era mValue


Nice! I like how this came out.

intricate · 2020-11-26T23:39:15Z

bors r+

2129: Initial CLI update for Allegra and Mary eras r=intricate a=dcoutts Re-implement the tx build-raw command in terms of the new era-generic TxBodyContent API. This should allow it to work for all eras, including Byron. It checks that each feature can be used in the selected era. There are still a number of TODOs, but this establishes the pattern. Co-authored-by: Duncan Coutts <duncan@well-typed.com> Co-authored-by: Jordan Millar <jordan.millar@iohk.io>

iohk-bors · 2020-11-27T00:21:57Z

Build failed:

ci/hydra-build:required

intricate · 2020-11-27T02:22:29Z

bors r+

iohk-bors · 2020-11-27T03:25:34Z

Build succeeded:

kevinhammond · 2020-11-27T13:55:33Z

Also, will the txout syntax need to be updated? Seems to only allow lovelace at the moment

intricate · 2020-11-27T15:15:51Z

@kevinhammond: Yup, there's a patch for that in #2072.

kevinhammond · 2020-11-27T17:53:26Z

Thanks!

dcoutts requested review from Jimbo4350 and intricate November 26, 2020 03:23

dcoutts requested review from erikd, mrBliss and newhoggy as code owners November 26, 2020 03:23

dcoutts force-pushed the dcoutts/cli-allegra-mary branch 2 times, most recently from 4f8b2c6 to 38700db Compare November 26, 2020 10:39

Jimbo4350 force-pushed the dcoutts/cli-allegra-mary branch 2 times, most recently from c4a53ca to 3a5591e Compare November 26, 2020 13:20

Jimbo4350 force-pushed the dcoutts/cli-allegra-mary branch from 3a5591e to f40abd6 Compare November 26, 2020 14:06

dcoutts force-pushed the dcoutts/cli-allegra-mary branch from b2089c1 to cb23382 Compare November 26, 2020 15:43

intricate mentioned this pull request Nov 26, 2020

WIP: Propagate cardano-api Allegra/Mary changes to cardano-cli #2112

Closed

dcoutts and others added 15 commits November 26, 2020 22:05

Make cardanoEraStyle a function of CardanoEra not a class method

857beb7

It's more convenient this way since it means we can use it when we have a CardanoEra value or when we have an IsCardanoEra class constraint.

Add a TestEquality instance for CardanoEra

f084a21

This lets us dynamically compare era tokens and establish if they're the same. This will be useful in the CLI for example where we read two files and need to check they're from the same era.

Add tx body feature conversion functions

83d1047

They tell us if the feature is available in a given era, and if so provide us with the evidence. This will make working with these features (e.g. in the CLI) easier to do in a era-general kind of way.

Export TxBodyError from API modules

e72ab7e

Update cardano-cli tx build command to use CardanoEra flag

115add2

Defaults to --shelley-era When we make a transaction we need to know which era it is for.

Add withCardanoEra function

9ccb163

For going from an untyped UseCardanoEra enumeration to the strongly typed CardanoEra era.

Rewrite runTxBuildRaw in terms of typed TxBodyContent API

d0854c9

We now use a per-field validation that checks that the feature is available to use in that era.

Add readFile* functions for reading in any era

e26ce1d

For Tx, TxBody and Witness. We need to be able to read these files at the types corresponding to any of the supported eras.

Generalise mkShelleyBootstrapWitness(es) over all (Shelley-based) eras

7734882

Easy because all the functions they call have already been generalised.

dcoutts force-pushed the dcoutts/cli-allegra-mary branch from cb23382 to 6b86804 Compare November 26, 2020 22:07

dcoutts added 2 commits November 26, 2020 22:56

Generalise tx submit command to all eras

081c690

Needs a slight tweak to submitTx in the API to cover the Allegra and Mary era cases.

Fix hlint warnings

9c9a10a

intricate reviewed Nov 26, 2020

View reviewed changes

intricate approved these changes Nov 26, 2020

View reviewed changes

Continue supporting --ttl option in "transaction build" command

a379bb7

intricate mentioned this pull request Nov 27, 2020

Initial lexer and parser for specifying Mary multi assets via the cli #2072

Merged

iohk-bors bot merged commit 62faef3 into master Nov 27, 2020

iohk-bors bot deleted the dcoutts/cli-allegra-mary branch November 27, 2020 03:25

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Initial CLI update for Allegra and Mary eras #2129

Initial CLI update for Allegra and Mary eras #2129

dcoutts commented Nov 26, 2020

dcoutts commented Nov 26, 2020

kevinhammond commented Nov 26, 2020

dcoutts commented Nov 26, 2020

intricate Nov 26, 2020

intricate Nov 26, 2020

dcoutts Nov 27, 2020

intricate Nov 26, 2020

intricate commented Nov 26, 2020

iohk-bors bot commented Nov 27, 2020

intricate commented Nov 27, 2020

iohk-bors bot commented Nov 27, 2020

kevinhammond commented Nov 27, 2020

intricate commented Nov 27, 2020 •

edited

Loading

kevinhammond commented Nov 27, 2020

		readValue :: String -> Either String Value
		readValue _maCliSyntax = Left "Need 2072 for MA cli syntax parser"

Initial CLI update for Allegra and Mary eras #2129

Initial CLI update for Allegra and Mary eras #2129

Conversation

dcoutts commented Nov 26, 2020

dcoutts commented Nov 26, 2020

kevinhammond commented Nov 26, 2020

dcoutts commented Nov 26, 2020

intricate Nov 26, 2020

Choose a reason for hiding this comment

intricate Nov 26, 2020

Choose a reason for hiding this comment

dcoutts Nov 27, 2020

Choose a reason for hiding this comment

intricate Nov 26, 2020

Choose a reason for hiding this comment

intricate commented Nov 26, 2020

iohk-bors bot commented Nov 27, 2020

intricate commented Nov 27, 2020

iohk-bors bot commented Nov 27, 2020

kevinhammond commented Nov 27, 2020

intricate commented Nov 27, 2020 • edited Loading

kevinhammond commented Nov 27, 2020

intricate commented Nov 27, 2020 •

edited

Loading