has joined #pact-scala-itv

has joined #pact-scala-itv

@david.j.smith you'll need to set up the slack integration for itv if you want to show the commits here

Hmmmm :thinking_face: Thanks I'll ponder that one :slightly_smiling_face:

It's not super important, don't stress if its politically too difficult.

has joined #pact-scala-itv

Gosh this is a hot topic at the moment :slightly_smiling_face: https://github.com/ITV/scala-pact/issues/98

This bug is apparently about basic auth too.

Although @harris.osserman you?ll need to speak to Ron about pact-jvm?s support for basic auth.

I think this one's for you @david.j.smith https://stackoverflow.com/questions/52033736/how-to-modify-the-target-folder-for-generated-pact-files

:scream: I?ll have a look.

https://stackoverflow.com/questions/52033736/how-to-modify-the-target-folder-for-generated-pact-files-while-doing-contract-dr/52050804#52050804 :fist:

has joined #pact-scala-itv

I?m a bit new to pact and contract testing might be a dummy question but; while writing a contract test for getting an entity `/resource/ID` when I use a random ID in my consumer and generate the pact file, now I need to make sure provider uses the same id to create this entity or is there a way to implement this without an actual value ?

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi @david.j.smith, first of all thank you for "scala-pact". I have a question regarding http://io.itv.com/scala-pact/roadmap.html

do you have any estimations of when Phase 5 - AMQP could be done

this is very crucial part

Hi @nikolajs.cmihuns, glad to hear you?re finding it useful. AMQP support is the most requested feature, it?s also very difficult! :slightly_smiling_face: There is a PR open around this that needs attention, but I don?t have an estimates right now I?m afraid. My time is very limited these days, and the scale of the problem requires a lot of coordination. It isn?t just about adding AMQP support but also about supporting Pact V3/V4 (which the open PR does not try to do).

ok, thank you for the answer

just quick one more question, scala-pact is completely separate from DIUS-pact-jvm family of libraries ?

has joined #pact-scala-itv

Yes it is.

has joined #pact-scala-itv

Hello, Is there a way to exclude tests when running `sbt pactTest`?

Without pact i would just do `sbt "testOnly -- -l TestExcludeTag"`

And i thought i could do something like `Test / testOptions += Tests.Argument("-l", "TestExceludeTag")`, but that prevents me from running the test at all (So i can't do `sbt "testOnly -- -n TestExcludeTag"`).

Figured out i could work around it by making a separate configuration and setting `Test / testOptions := Seq(Tests.Argument("-l", "TestExcludeTag"), AllTestConfig / testOptions := Seq.empty` if anyone else was wondering.

@david.j.smith is there any plans to implement tagging releases at publish / using tags to select which version(s) of the pact to verify? Would you accept pull-requests?

Hi @jonas.natten - sorry for the slow response - very busy at the moment. I don?t really do test exclusion myself, so it isn?t something I?ve looked at hard. But it?s just ScalaTest, so as long as your consumer tests are split up appropriately I assume the usual processes will apply? The thing currently in progress - nearly there - is adding metadata support as a step towards Pact V3/4. I absolutely do take PR?s. I strongly advise that unless the issue is trivial, you should raise an issue explaining the problem and how you?d like to solve it, let me feedback, and then do the PR so it?s not a total shock. I desperately want to merge in all PR?s and I try hard to do so, its hard than it seems. People make it hard by either a) just turning up with a code change and expecting me to take their word for it, or b) discussing it first but then arriving with 2000 lines of changes that touch a massive area of code? and expecting me to take their word for it that it?s all still ok?. if I have to take two days off work to wade through it all, it probably just won?t get merged! :shrug: :slightly_smiling_face:

Sounds good @david.j.smith. I updated the issue i created asking here: https://github.com/ITV/scala-pact/issues/107 :slightly_smiling_face:

It?s a great PR! I left feedback - thanks! :slightly_smiling_face:

Thanks for merging the tag PR @david.j.smith. I've made a second one for the missing url encoding here: https://github.com/ITV/scala-pact/pull/115

Also implemented http basic auth: https://github.com/ITV/scala-pact/pull/114 Not sure if its implemented in a stupid way, but needed it badly for a project so i just went ahead and did it :slightly_smiling_face:

A bunch of lovely people have submitted PRs during December. They?re mostly merged but I?m trying to get to the bottom of a bug in the Http4s `0.20.0-M4` module. I think I?m going to raise an issue on the Http4s project soon and just get the next version of Scala-Pact out with a small disclaimer about chunked requests not being accepted yet.

@jonas.natten Running the testing scripts now. If all is well I?ll try and cut the release too. Thanks for all the work. :smile:

Sounds good! I ran the scripts/local-build-test.sh script, but please verify! :+1:

Ah thanks, that?s mostly what I?m doing :wink:

Looking good. I?ll start the release process and let you know when it?s done - it takes a long time?

Well. In theory `2.3.4` is now out and should be resolvable via SBT in the next 15 minutes (takes a while to propagate). It should then appear on http://search.maven.org in the next couple of hours.

@jonas.natten http://io.itv.com/scala-pact/change-log.html#2-3-4-2018-12-31

Woho! :tada:

Thanks for the help and happy new year! :smile:

has joined #pact-scala-itv

I can't seem to find a way to provide a map in provider state in scala-pact. Such a functionality is supported in pact-jvm. Can someone tell if they don't find it necessary or if it will be supported in near future

Hi @nishantv12, this is a quiet channel - helps if you @ me. :slightly_smiling_face: Can you give me an example?

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi Everyone, I am getting the following error writing provider tests and couldn?t figure out why maybe somebody has some idea.

I need to find a better notification system for this Slack channel :sweat_smile:

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi Team

Do we support message queue based contract validations in scala (pact spec v3)?

Not currently, it?s an incredibly slow burn work in progress?.

Thanks @david.j.smith

Hey all, I was trying to use pact jvm library for java/junit to validate scala provider method

has anyone tried this before?

I am facing some problem with compiling both maven and sbt, somehow maven doesn't honor dependencies defined in sbt file and throws exception

I'd try the #pact-jvm channel @kashikar.harsh

Sure @david.j.smith, will post it there as well

This channel is for a different implementation entirely, that just happens to also be for Scala. :slightly_smiling_face:

Thanks, will post my queries to pact-jvm hereafter :+1:

@amber.race has left the channel

Hey all

I am trying to parameterize pact broker credentials by using variable pactBrokerCredentials as stated in doc https://github.com/ITV/scala-pact/blob/master/scalapact-docs/src/main/paradox/basics/sbt-commands.md

however getting an error that pactBrokerCredentials cannot be resolved

using SBT 0.13.16 and scalapact 2.2.5

Hi @kashikar.harsh. Looking at the repo at the point this sbt setting was added, the version was around 2.3.4 - i.e. https://github.com/ITV/scala-pact/blob/2d4f220d01a0c2d86f8aaaeff2ef8d2b06bc56d8/build.sbt#L64 Looks like you'll need an upgrade :smile:

Thanks @daniel.tattan.birch

Is there an example where we pass pactbroker credentials as argument from sbt?

You can pass basic auth in the url if that helps?

We do log urls and responses for troubleshooting, so we will end up logging the credentials in plain text with that approach

I?d need to check but I?m pretty sure it gets converted to the standard auth header - do you also log headers? And you?re logging access requests to Pact Broker?

we have a common method for all http requests that logs the url information

Even during tests?

Getting back to your question, a better tactic than supplying command line arguments is to have a `.gitignore`d `pactcredentials.sbt` file in your root directory that declares the `pactBrokerCredentials := ("username", "password")` line (http://io.itv.com/scala-pact/basics/sbt-commands.html#command-line-options).

However! Pact _is_ going to use those credentials to make a basic auth header.

That?s just how it works.

has joined #pact-scala-itv

Is there a way to publish contracts without running the tests. So basically I have generated my pact files and now just want to publish. However `pactPublish` command runs all the tests

Yes

There are SBT Commands and SBT Tasks.

When I wrote the commands (the tasks came much later) I was trying to make the process robust by ensure everything, like `pactPublish` ran from a clean slate so that you could never get into a bad state.

However, Commands don?t compose and most people who?ve been using SBT for decent period of time are used to dealing with incremental build state, so we introduced Tasks that do only the intended work, but offer no guarantees about accuracy.

You can use `pactPush` to do the thing you want, you just need to be aware that it will push _everything_ even old, lingering, or uncompiled pact files. Suggest running `pactPack` then `packPush` to flatten pact files together and then push them.

Thank you very much !

Is there a way to pass parameters to either pactPush or pactPack in a multimodule project ?

That?s an excellent question. :thinking_face:

Example?

I get this error basically `Supplied pact path does not exist! Aborting file load.`

Oh I see.

Or do I need to define this path somewhere because in my multi-module project there is only one module which has contract tests now

And you?re doing something like `sbt mymodule/pactPush`?

no just doing it at the highest level

I get the same error doing it with the submodule/pactPush

The opposite of what I thought then. You run at the highest level, the pact files are generated at the module level but you try and publish from the root?

Just seeing if we do this anywhere

Files are generated at the highest level then I tried to push from both the highest level and from the submodule

It turns out I?ve already been pushing the contracts to the broker even though I got errors

Yes I?ve just had a similar issue, I was testing it out on one of our projects.

In a multi project build, pact allows works in the root level. However, your best bet is to always invoked the commands at the module level especially if you?re aggregating projects because otherwise it runs the command once per aggregated project, even though a previous pass might have deleted the files or something.

Multi project builds through up loads of corner cases!

(For plugin creators)

Another question: ```"metadata" : { "pactSpecification" : { "version" : "2.0.0" }, "scala-pact" : { "version" : "2.3.12" } }``` How do I add this block to my contracts the sample project generates them but couldn?t find anything different that will add these

I just found the issue on github and I guess this has been fixed but I am using the older version so there is not much I can do i suppose ?

That?s right, we added this in a fairly recent version, it just does it automatically now.

Just to double check the README file vs the official documentation page the latest versions seems to be different; is there a reason or just the docs need to be updated

Oh heck - sorry the latest version is `2.3.12` - sorry about that!

No worries just wanted to let you know; thank you for the update

Thanks for pointing it out! :slightly_smiling_face:

```[error] Test suite contracttests.ConsumerSpec failed with java.lang.NoClassDefFoundError: io/circe/generic/encoding/ReprObjectEncoder. [error] This may be due to the ClassLoaderLayeringStrategy (Flat) used by your task. [error] To improve performance and reduce memory, sbt attempts to cache the class loaders used to load the project dependencies. [error] The project class files are loaded in a separate class loader that is created for each test run. [error] The test class loader accesses the project dependency classes using the cached project dependency classloader. [error] With this approach, class loading may fail under the following conditions: [error]  [error]  * Dependencies use reflection to access classes in your project's classpath. [error]    Java serialization/deserialization may cause this. [error]  * An open package is accessed across layers. If the project's classes access or extend [error]    jvm package private classes defined in a project dependency, it may cause an IllegalAccessError [error]    because the jvm enforces package private at the classloader level. [error]  [error] These issues, along with others that were not enumerated above, may be resolved by changing the class loader layering strategy. [error] The Flat and ScalaLibrary strategies bundle the full project classpath in the same class loader. [error] To use one of these strategies, set the  ClassLoaderLayeringStrategy key [error] in your configuration, for example: [error]  [error] set Test / classLoaderLayeringStrategy := ClassLoaderLayeringStrategy.ScalaLibrary [error] set Test / classLoaderLayeringStrategy := ClassLoaderLayeringStrategy.Flat``` I even set these strategies but still getting errors

Can you privately send me your sbt dependency lines (copy+paste)?

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi, has any one experience with Kafka message CDC testing? I'm trying, though I got stuck. The scalapact library seems only to support HTTP call testing.

Yes, I believe this is the case.

Scala pact hasn't had much love for a while.

Pact-JVM supports message queues - I don't know if you're able to use that interface?

(I'm not a Scala dev)

I tried to, byut didn't have any success yet.

Sad but true, just maintenance.

We?re all still using it every single day, since for the HTTP use case it works very well. But no-one has felt sufficiently moved to carry on the messaging protocol work as yet, and I?m just too busy these days.

I can sympathise @david.j.smith!

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi @david.j.smith :slightly_smiling_face:

Oh hello! :smile: @naree.song

I found the right slack in the end. I am playing with `scalapact` and it seems to be working as well as I remember :slightly_smiling_face:

This is like a collision of worlds or something.

I apologise then.

:wink:

Oh I messaged you on linkedin as I was looking for slack channel or gitter channel for `scalapact`

Oh sorry! I hardly ever venture there.

no worries :slightly_smiling_face:

This is the only thing that exists and it?s pretty quiet as you can see. :slightly_smiling_face:

Hahaha, quiet is a good thing sometimes :slightly_smiling_face:. I am planning to use `scalapact` for the API project I am starting on at work. The only thing that got me slightly lost was that I could not find something like compatibility matrix to find all the dependencies I need to use `scalapact` with Scala 2.13

I am wondering if you guys are planning to add it to the documentation? If not, I am happy to do it.

Good question.

We haven?t updated that for quite a while.

Just a sec

(Noel says hi :slightly_smiling_face: )

Hi to Noel :wink:

Looks like the main things to watch out for at that we?re only publishing 2.13 support for: http4s 021 circe 12, 13

(I just did a quick delve through the build file)

There are some other people here helping keep it up to date (Do you know Gus? Andrew Gustafson? He?s one anyway.)

It?s still massively in use so we know it?s ok, just out of date with regards to the pact spec.

Cool, yeah personally I think pact testing is the magic ingredient that enables the microservice architecture :slightly_smiling_face: I really wanted to use it for my previous project but it was Kafka. I saw someone asking about it here as well but if I were to work on another Kafka project, I would like pact support for Kafka. Happy to work on it myself :slightly_smiling_face:

In the meantime, I am able to use `scalapact` with Scala 2.13 nicely for restful endpoints. :slightly_smiling_face:

Can the documentation on `scalapact` on github be updated to include the compatibility matrix? I am happy to do it and create a PR for it if that helps?

Yes I know him.

All help is gratefully received, I don?t personally spend much time on it these days but I?m going to hopefully rope someone into working on it later in the year.

The main thing missing is message support that would meet a RabbitMQ or Kafka like use case. Other variations of Pact do support this.

I haven?t tried it in years, but there is Pact-JVM that might do what you want?

But yes if you?d like to help with docs, or working on messaging we?d be delighted to talk through what is required to make it happen - we have started, I think there might be open issues around it.

Cool, I will look for the issue and if I cannot find it, I will create one and we can continue the discussion there. Thanks Dave :slightly_smiling_face:

Lovely, thanks Naree! :smile:

This sounds ominous :joy:

Ha! I didn't necessarily mean you Dan. :wink::laughing:

has joined #pact-scala-itv

has joined #pact-scala-itv

Hi, we are getting ```[scala-pact] 1 Pact verify tests failed. Publish verification results failed with 400``` on attempt to publish verification result to broker, which doesn?t use authorization. Is anybody able to give a hint with it?

It?s been a long time since I?ve looked at that corner of Scala-Pact, but `400` I guess its a bad request status code from Pact Broker. Anyway to manually make the call and see if you and recreate the problem?

Will try then, thanks

It?s quite possible that the verification results API spec has changed and we?re out of date. We use a standalone pact broker here, but I doubt it?s the latest version.

We are using internal verification for what it?s worth.

For what it?s worth the broker responds with ```{ "errors": { "success": [ "must be boolean" ] } }``` to request that goes internally.

My colleague has opened a pull request with a fix: https://github.com/ITV/scala-pact/pull/163

Cc: @bethskurrie Time for pact tests?

has joined #pact-scala-itv

Hi there! My name is HeyTaco!, and you can use me to give people tacos to show your appreciation. My tacos will spread joy through Slack!

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

how do you publish verification results when using `verifyPact` syntax in tests?

hm, maybe `ScalaPactSettings.publishResultsEnabled` ? documentation really needs to be updated

has joined #pact-scala-itv

has joined #pact-scala-itv

> By default, Scala-Pact does not allow you to publish pact files from SNAPSHOT versions of your project (but takes into account the pactContractVersion if set). This is because it can confuse pact brokers understanding of the latest contract.

can anyone elaborate on this?

i read https://docs.pact.io/getting_started/versioning_in_the_pact_broker which provided a little clarity but i?m still not sure why SNAPSHOT versions would ?confuse the brokers understanding of the latest contract?

especially because the pacts would be tagged, so i don?t see the issue

It may have been historic due to versioning schemes in the broker

pretty sure i have something misconfigured somewhere, but I am using `pactBrokerWithVersionSelectors` with `publishResultsEnabled` and everything is working correctly except the provider version is not getting tagged with the branch like the consumer is?

i think this might be a bug in `scala-pact`, filed https://github.com/ITV/scala-pact/issues/207

Yes that, also there was no concept of ?ready to deploy? back in the mists of time when I wrote that. :slightly_smiling_face:

Thanks - we?re not terribly good at checking this channel! :slightly_smiling_face: I?ll try and encourage others who help maintain Scala-Pact to come along.

i?ll take a look at how the consumer does this and see if i can perhaps open a PR to fix this

would be great! our company mainly uses scala and we are just now adopting pact, so hopefully some of us can help contribute back to this project

has joined #pact-scala-itv

started a thread in #pact-jvm about scala-pact vs pact-jvm

has joined #pact-scala-itv

Hello Folks, Where could I get reading material & kickstart examples with explanation, for someone new to Scala & ITV_pact? Thanks

@rafael.anachoreta has left the channel

@jackbwheatley i see some new commits to the project which reference pact-jvm ? can you shed any light on what you?re working on?

has joined #pact-scala-itv

we are doing a proof of concept right now for pact v3 message support in our scala projects, and we are bypassing the scala-pact project because it?s too much work to do it there

the plan is to integrate it directly on top of pact-jvm

part of that includes publishing pact files generated from pact-jvm, which we would need to write support for probably into an sbt task

Hi Kyle, so I?ve actually been having some similar thoughts about scala-pact. As you allude to, adding v3 features of the pact spec to scala-pact would be an inordinate amount of work. As an experiment, I?ve been quietly working on a pact-jvm wrapper - https://github.com/jbwheatley/pact4s - and we are considering phasing out our use of scala-pact at ITV in favour of wrapping pact-jvm and the using the broker CLI. Part of that roadmap is ensuring that we can continue to use scala-pact?s sbt plugin to publish and retrieve pacts - hence the release earlier this week that adds support for publishing pacts made using pact-jvm.

thanks for the update ? looks like support for message testing is still in the works, but it has request/response testing already, is that right?

i do see `MessagePactForger` which i guess is for the consumer tests. it doesn?t seem like there is any relation between a message and a test, i guess it?s up to the implementor to ensure every message is tested?

Yep, its still early days, and I need to start documenting things. The test classes are a good place to start for examples. I have been able to write consumer http pacts with pact4s and publish them to the broker with the scala-pact sbt plugin in one of our services, so I can at least say its functional in that regard :grin: > i guess it?s up to the implementor to ensure every message is tested? yeah that is right.

am i correct that we can?t yet create and verify message pacts with this library yet?

i guess perhaps create but not verify

it looks like `runVerificationForConsumer` is only for interactions (not messages)

yeah that?s right, I haven?t looked into message pact verification yet

oh, actually `runVerificationForConsumer` looks like it does support message pacts

oh right, is there something in the `ProviderInfo` or version selectors that says they are message pacts?

yeah i guess `Interaction` has multiple sub-types now, one of which is `MessageInteraction`

although that validation happens in `verifyResponseByInvokingProviderMethods` which seems to require test method annotations

(i?m looking at https://github.com/pact-foundation/pact-jvm/blob/master/provider/src/main/kotlin/au/com/dius/pact/provider/ProviderVerifier.kt)

`runVerificationForConsumer` calls `verifyInteraction` which ends up here https://github.com/pact-foundation/pact-jvm/blob/master/provider/src/main/kotlin/au/com/dius/pact/provider/ProviderVerifier.kt#L512

ok seems like this needs some investigation. Would rather not have to use annotations

agreed

pact-jvm is a little hard to reason about sometimes :sweat_smile:

for sure, its been a bit of a minefield for me so far

has joined #pact-scala-itv

hi! anyone having issues with `sbt pactVerify` ? for me, it returns timedout broker, still curl from logs works flawlessly

more info is here - https://pact-foundation.slack.com/archives/C9VPNUJR2/p1622152897049400

i don?t use `sbt pactVerify` (we do verification from Scalatest), but we haven?t had any issues with the library communicating with the broker

our client timeout is set to 2 seconds

do you have producer and consumer in same repo ?

@kflo do you verify with `sbt test` ? how then producer sends verification to broker ( if you use it) ?

found out - extra param in ```pactBrokerUseLatest```

hello once again :slightly_smiling_face: for provider one can set such pact sources ? loadFromLocal ? pactAsJsonString ? pactBrokerUseLatest latter doesn't use params from `pact.sbt` is it intended, or I didn't found how not to duplicate all those in each test ? ``` case class pactBrokerUseLatest( url: String, provider: String, consumers: List[String], publishResultsEnabled: Option[BrokerPublishData], pactBrokerAuthorization: Option[PactBrokerAuthorization], pactBrokerClientTimeout: Option[Duration] )```

correct, we use `sbt test` and configure it to report the results back

hello maybe one more question there is example about https://github.com/pact-foundation/pact-js/blob/master/examples/messages/consumer/message-consumer.spec.ts any examples about producer/consumer tests for ws in scala ?

are you referring to websockets/messages? the itv/scala-pact project doesn?t support messages yet, it only supports V2 (message support was added in V3). as you?ve discovered we have been working on a new library that is built on top of `pact-jvm` here https://github.com/jbwheatley/pact4s ? that one will support message testing

has joined #pact-scala-itv

has joined #pact-scala-itv

Hey guys, does scala-pact supports messaging pact?

@varnit.garg2424 :point_up:

Hi @varnit.garg2424 - it does not no. Scala-Pact is slowely being replaced by https://github.com/jbwheatley/pact4s - @jackbwheatley I guess pact4s would support messaging?

Yes, pact4s does support message pacts! It would great to get more people to try it out :simple_smile:

Great! Thanks @jackbwheatley. Let me try it out. By the way, do we have any sample project for reference to get started with? :slightly_smiling_face:

@varnit.garg2424 the message pact work hasn't been documented yet, but you can find examples on both forging and verifying pacts in the tests suites

i wonder if we should create a new channel for pact4s :thinking_face:

or maybe just #pact-scala

@jackbwheatley what do you think?

this one is also about scala

so you can create #pact-scala and mention 2 branches there :smile:

haha. this existing channel is specifically for https://github.com/ITV/scala-pact

i?m wondering if it?s better to just make an all-purpose #pact-scala instead of tying it to a specific library

in any case, we would need the blessing of @matt.fellows or @bethskurrie to actually create the channel

hi! anyone succeeded to create `pact4s` scala message consumer ? i?m having troubles with that ( copypasted example https://github.com/jbwheatley/pact4s/blob/main/scalatest-pact/src/test/scala/pact4s/scalatest/MessagePactForgerScalaTestSuite.scala, but no luck ) ```An exception or error caused a run to abort: 'void io.circe.Codec.$init$(io.circe.Codec)' java.lang.NoSuchMethodError: 'void io.circe.Codec.$init$(io.circe.Codec)' at pact4s.circe.implicits$$anon$1.<init>(implicits.scala:38) at pact4s.circe.implicits$.<clinit>(implicits.scala:38) at com.xx.yy.api.MessagesConsumerPactVerifierSpec.<init>(MessagesConsumerPactVerifierSpec.scala:21) at java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) at java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:64) at java.base/jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45) at java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:500) at java.base/java.lang.reflect.ReflectAccess.newInstance(ReflectAccess.java:128) at java.base/jdk.internal.reflect.ReflectionFactory.newInstance(ReflectionFactory.java:350) at java.base/java.lang.Class.newInstance(Class.java:645) at org.scalatest.tools.Runner$.genSuiteConfig(Runner.scala:1402) at org.scalatest.tools.Runner$.$anonfun$doRunRunRunDaDoRunRun$8(Runner.scala:1199) at scala.collection.immutable.List.map(List.scala:246)``` for sure its something simple :)

@jackbwheatley maybe? ( sorry for pinging )

not sure about that one! What version of circe are you using? NoSuchMethodError would suggest some sort of binary compat issue..

> val circeVersion = ?0.13.0? > val enumeratumCirceVersion = ?1.6.1"

ah right, pact4s-circe is using circe 0.14 - perhaps try bumping that dependency?

heh, ok, will try

as a note

thing didn?t worked - `import pact4s.circe.implicits._` it doesn?t fetched as dependency, so, i?ve added explicitly > val pact4sCirce                 = ?io.github.jbwheatley? %% ?pact4s-circe?                             % ?0.0.10? but this one fetched automatically > io.github.jbwheatley:shared_2.13:0.0.10:jar

yes, that is intentional as not everyone uses circe. I think there is a note buried in the readme that you need the additional dependency, but it could be more clear.

yep

indeed

> By adding the additional dependency `io.github.jbwheatley %% pact4s-circe % xxx`, instances of `pact4s.PactBodyEncoder` are provided for any type that has a `circe.Encoder` instance. This allows the following:

will try bumping :+1: thank you

`pact4s` guys, is this valid ?

like 2 messages with same annotation

or there are other options form that ?

yes, bumping circe from 0.13 to 0.14.1 done magic ( cc @kflo)

what is the intent? maybe you want to use a provider state instead?

so consumer has ``` "$.mode": { "matchers": [ { "match": "regex", "regex": "^(Live|Maintenance)$" } ] },```

and producer makes sure, all owns states are fed

yeah seems like a good use-case for a state. but also maybe a test that can be covered elsewhere outside of pact

seems like it might be testing business logic

yes and no

how consumer will know, that `Maintenance` is changed to `Hidden`

producer feeds all own stuff and suffers

this might be a better question in general about how to test things in pact

doesn?t seem specific to pact4s

did that already :slightly_smiling_face:

but maybe scala power has own answers already :slightly_smiling_face:

i will say that, if pact-jvm supports something (like multiple `@PactVerifyProvider()` with the same description), then pact4s will also support it

it works, yes

but it seems like the primary question here is what the best pattern is

but question is more general

> is what the best pattern is exactly

i unfortunately haven?t spent enough time implementing pacts to answer it

as growing permutations will not scale with copypaste

yeah, a sign that maybe it?s not a good pattern

:stuck_out_tongue:

i feel like using states is the best pattern. but that will mean there could be other states that the provider supports that the consumer doesn?t

like if the provider added `Down`, using states wouldn?t alert you to that

because you are only requesting the states you handle, as a consumer

exactly

yes, but it is not that case, imho

it is a murky area https://stackoverflow.com/questions/43089653/should-i-test-all-enum-values-in-a-contract

otherwise one can change `"regex": "^(Live|Maintenance)$"` to simple `like`

awesome link with properly asked question :taco:

@kflo :taco:

haha

and Matt says ?NO? :man-shrugging:

i guess it ultimately comes down to what you expect your consumer to do if it is given a different enum value than it knows about

> there is no right or wrong answer, just be wary of testing _all_ permutations of input/output data. yeah

looks like we need to sit down and discuss, which road we will go :)

it?s kind of weird, but i wonder if the consumer could expect the producer to return a field that is the combination of all the enum types, something like `Down,Live,Maintenance` ? it would have to be sorted. but then the consumer could expect a specific list, and the provider build would fail if it has a different value. the list would have to be generated directly from the enum, otherwise it would be useless

seems like a code smell though

:smile:

it will be a bit different test

still, it could be another pact

yep, i was thinking it would be a separate pact

enum vs enum

probably an HTTP pact

yep, like all states :+1:

with exact match

from one other view - its code changes for testing

like adding extra endpoint

yep

so, tomorrow is Friday - nice time for holywars :smiling_imp:

lol

hi! any ideas about this ? ```MessagesConsumerPactForgerSpec *** ABORTED *** [info] java.lang.Exception: Auto-conversion of circe.Json to PactDslJsonBody is unsupported for json containing arrays. [info] at pact4s.circe.JsonConversion$.$anonfun$addFieldToBuilder$7(JsonConversion.scala:33) [info] at io.circe.Json.fold(Json.scala:39) [info] at pact4s.circe.JsonConversion$.addFieldToBuilder(JsonConversion.scala:34) [info] at pact4s.circe.JsonConversion$.$anonfun$_jsonToPactDslJsonBody$4(JsonConversion.scala:48) [info] at scala.collection.IterableOnceOps.foldLeft(IterableOnce.scala:646) [info] at scala.collection.IterableOnceOps.foldLeft$(IterableOnce.scala:642) [info] at scala.collection.AbstractIterable.foldLeft(Iterable.scala:919) [info] at pact4s.circe.JsonConversion$.$anonfun$_jsonToPactDslJsonBody$3(JsonConversion.scala:47) [info] at io.circe.Json.arrayOrObject(Json.scala:56) [info] at pact4s.circe.JsonConversion$._jsonToPactDslJsonBody(JsonConversion.scala:46)```

so its failing for ```{ "options" : ["a","b"] }```

@jackbwheatley will have to answer that one

but as it says, can?t use arrays when converting json body with circe. i don?t know why though

Its because I think you can only do structural matches on array items in pact-jvm, rather than direct comparisons. I used this documentation as a guide when writing the `json -> PactDslJsonBody` converter: https://docs.pact.io/implementation_guides/jvm/consumer#building-json-bodies-with-pactdsljsonbody-dsl If you think you can figure out a way to do it, would be great to see!

maybe it can be done with this in v4 of the pact spec? https://docs.pact.io/implementation_guides/jvm/consumer#array-contains-matcher-v4-specification

ok, will check tomorrow :+1:

Dmitry - do you have a provider that always returs back a list of objects of different types/shapes?

The array contains matcher was created for the hypermedia use case, where it?s possible to get back links with different shapes and values, but where it?s really important for the consumer to be able to rely on a particular link to be there

it also works well when you essentially want to ?pluck? objects from a heterogenous set of objects in a list

Without the array contains matcher, if your consumer says it needs ?an array of <specific object/type>? but provider has `[<your type>, <some other type>, <your type>, "<another different type>]` it won?t work.

it is message with one of the fields as an array of strings

an array of strings should be supported, if : 1. The matcher accepts the format of each of the strings 2. It?s an exact match

can you share what your expectation looks like an the actual from the provider?

ah, forgot to mention - its a consumer part

@jackbwheatley looks like https://docs.pact.io/implementation_guides/jvm/consumer/#array-contains-matcher-v4-specification is not right thing - we wanted to feed `withContent` with real objects, not forged, to have most strict checks

having a look at what we can do around arrays now

@dimundo new version that should work with arrays coming shortly

`v0.0.12` has hit maven now

Awesome, it works, thank you @jackbwheatley :taco:

@jackbwheatley is it correct one in ProviderInfoBuilder? > includeWipPactsSince: Option[FiniteDuration] = None, as it expects something like `7 days`, not `2021-06-12` from https://docs.pact.io/pact_broker/advanced_topics/wip_pacts/

The finite duration is time since the epoch, it gets converted to a date to pass to pact-jvm

so, if i need pacts 7 days old ? i?m not so good at scala as i wanted :(

If you have some other way of creating a java date or instant, you should be able to convert that to seconds or millis, then convert that to a finite duration

e.g. ```import scala.concurrent.duration._ val date: java.util.Date = ??? val instant: java.time.Instant = ??? date.getTime.millis : FiniteDuration instant.toEpochMillis.millis : FiniteDuration```

something like this ?

that would work, but you can also do: ```import scala.concurrent.duration._ System.currentTimeMillis.millis - 7.days```

oh, nice

thank you :+1: @jackbwheatley :taco:

np :simple_smile:

it?ll be easier once I get around to adding a scaladoc for stuff like this :grin:

```Some(new FiniteDuration(System.currentTimeMillis() - 7.days.toMillis, MILLISECONDS))``` `.toMillis` was missing :slightly_smiling_face: ( just for the record, is someone will search this :slightly_smiling_face: )(

i have a feeling, that WIP doesnt work :disappointed:

if i define Provide like this

it verifies only `cit1` tagged versions

if i change to `selectors = ConsumerVersionSelector(None) :: Nil` aka no tags - each version is verified

but in such case, latest consumer version could be non compatible

cc @kflo maybe ?

i always use consumer with latest tag (aka not providing any tags)

but what is latest ?

actually it uses the branch tag for the consumer at latest

iirc, latest is latest published, not latest verified, no?

and if there is no matching branch ?

if there is no matching branch, it will not find any pacts for that consumer

but you should always be tagging your pacts

```Find all the pacts that are the latest for their tag (the "head" pacts). (eg. The latest pact for a version with tag main + the latest pact for a version with tag feat-x + the latest pact for a version with tag feat-y ...) Discard the pacts that have been explicitly selected in the consumerVersionSelectors. (eg. usually you'd configure the consumer version selectors with {"tag": "main", "latest": "true"}, so discard the latest main pact).``` so, this is exactly what should happen - all latest consumers , except `cit1` ?

we don?t use WIP pacts in CI, only in CD. we found it too confusing to use in CI

can you explain ?

i tagged you on another thread, but basically we ended up going with this workflow https://docs.pact.io/provider/recommended_configuration#verification-triggered-by-provider-change

which seems like the default workflow most people will want

related: https://github.com/pact-foundation/pact_broker/discussions/391

so, i will post kinda workflow in thread

CommonAPI provider having `wip` turned on, from own branch `feature/BK-335-pact-wip-debug` / version `a31ac102` verified earlier changed consumer pact tag `OPTO-58` / ver `5dd38d1c`

provider merged to master and changed to `eb30a482` / `master` , still consumer is ok

after deploy > Tagging CommonAPI version `eb30a482` as `cit1` so far so good

and i did that for wrong client :facepalm:

ah, but this is the catch

https://docs.pact.io/pact_broker/advanced_topics/wip_pacts/ > Find all the pacts that are the latest for their tag (the ?head? pacts). so, if many consumers has same tag, only 1 consumer will be selected

so , latest tag `tag1` is only `consumerA` and `consumerB` is deselected from WIP pacts

@kflo, looks like it is solved :smile: but not sure, how to proceed

> so, if many consumers has same tag, only 1 consumer will be selected it should be the latest version of each consumer pact with the tag is selected

for example, all consumer pacts tagged `main` or `production`

at their latest versions

aint latest tag is broker-wise?

what is latest `main` for you ?

all the consumer pacts where the code has been merged to `main`

it gets tagged with `main` by the CD system, which is triggered by code merging to `main`

but if you open pact broker and choose `Show latest tags`

how many tags `main` will be there ?

one for each consumer, assuming the code was merged

i don?t know where ?show latest tags? is

i see only each tag as unique

the tag is unique, but it is applied to each consumer pact

there are many `tag: main` , weird

mine are unique :disappointed:

> *x-pact-broker-version:* 2.80.0

how are you publishing your pacts?

ok so you are tagging them on publish, that should work. we do the same (just through the SBT plugin right now)

thats from JS part, but this shouldn?t matter

so, for reproduction there is a need in more ~pervert~ situation due to having monorepo for consumer ?  `consumerA` with changes in contract has  `version1` and `tag1` and contract with `ProviderC` ?  `consumerA` with changes in contract has  `version1` and `tag1` and contract with `ProviderD` ?  `consumerB` with changes in contract has  `version1` and `tag1` and contract with `ProviderC` ?  `consumerB` with changes in contract has  `version1` and `tag1` and contract with `ProviderD` consumers are still in branch=tag= `tag1` and WIP tests against master with tag= `tag2` are skipping `consumerB` WIP verification

some mindblowing, but ¯\_(?)_/¯

hm, not sure i fully understand. we also are using a monorepo for all our services (consumers and providers). where are the WIP tests being run from? the main branch? remember that if a provider verifies a contract with the same content at any time, it will no longer be considered WIP

started with trunk of producers, but then all-in, as provider/provider tests not touched for some time

has joined #pact-scala-itv

https://github.com/pact-foundation/pact_broker/pull/469 looks like this was the cause of havin too small latest tags - https://github.com/pact-foundation/pact_broker/pull/469

has joined #pact-scala-itv

@jackbwheatley we are currently using `scala-pact` for HTTP pacts and `pact4s` for message pacts. they don?t seem to play well together right now, because `scala-pact` requires ?squashing? the HTTP pact files, but `pact4s` does not. I know that the latest version of `scala-pact` allows you to specify that you are using only `scala-pact` files or only `pact4s` files, but doesn?t seem to work for both. it would be most desirable if we could just run `pactPack` only for the `scala-pact` files and have it just ignore the pact-jvm ones? what is the recommendation? the only thing i can think of is putting the two into separate folders and moving them around before publishing, which is a bit of extra complexity i?d like to avoid

we do plan to eventually switch to `pact4s` for everything

i guess another solution is running just the `scala-pact` tests, then `pactPack`, and then running the message tests. that will require some extra configuration in the build to separate them though

probably will have to do that for now

actually it seems pretty trivial to switch over our request/response pacts to `pact4s`, so i?m going to suggest we just do that :slightly_smiling_face:

hey @jackbwheatley @matt.fellows @bethskurrie i think `pact4s` is getting enough traction now that we should have a separate channel for it

https://github.com/jbwheatley/pact4s <--

i would love if it was eventually part of pact-foundation, but we can discuss that later

(i ask because it seems like i can only create private channels, which maybe i should do for now?)

Done!

thanks @matt.fellows!

good news, we now have a separate channel just for `pact4s`: #pact4s

has joined #pact-scala-itv

hi there. i've recently started seeing the scala pact library fail when trying to decode a pact from pactflow. it looks as though it gets to this point of the `PactImplicits.scalaPactDecoder`, but then fails with the below error ```_links <- cur.downField("_links").downField("curies").http://delete.as[Option[Links]]``` `DecodingFailure([A]Option[A], List(DownField(title), DownField(pb:consumer-versions), DeleteGoParent, DownField(curies), DownField(_links)))` which becomes `Could not read $dataType from json: $json` this started happening without any dependency changes, so i suspect something has changed in how pactflow sends its pacts as JSON. curiously, if i try to run the verification locally with the same pact JSON, it fails unless i delete one of the following from the `_links` section of the pact - if `curies` or `pb:consumer-versions` is removed, the pact succeeds. ```// MUST REMOVE "curies": [ { "name": "pb", "href": "https://<companyName>.http://pact.dius.com.au/doc/{rel}?context=pact", "templated": true } ], // OR REMOVE "pb:consumer-versions": [ { "title": "Consumer version", "name": "<version>", "href": "https://<companyName>.http://pact.dius.com.au/pacticipants/<service>/versions/<version>" } ],```

i've seen this issue has been raised on GitHub which looks to be related: https://github.com/ITV/scala-pact/issues/231

i've also tried bumping the scala-pact version to latest, but this was fruitless

further discussion on the #pactflow channel https://pact-foundation.slack.com/archives/CLS16AVEE/p1635324284042200

just to follow on from the above, this has been fixed in `scala-pact:4.3.0` - thanks @jackbwheatley for getting this resolved so quickly!

has joined #pact-scala-itv

Hi Folks, I?m new to pact testing and I?m having the following issue, I would really appreciate any insights you have. Besides we have not included any changes on the pact tests recently, our CI pipe has started to fail, I leave some of the logs below.

`Error: Could not read scala-pact pact from json`

`Could not convert good response to Pact:`

[info] java.lang.Exception: Failed to load consumer pact from: ***/pacts/provider/project/consumer/offer/pact-version/5d521cb94a9ce7a42c96067b52cf5ddaeff02368/metadata/c1tdW3RdPW1hc3RlciZzW11bbF09dHJ1ZSZzW11bY3ZdPTEyMDgmc1tdW3RdPXByb2Qmc1tdW2xdPXRydWUmc1tdW2N2XT0xMTcyJnNbXVt0XT10ZXN0aW5nJnNbXVtsXT10cnVlJnNbXVtjdl09MTIwOA [info] at com.itv.scalapactcore.common.PactBrokerClient.$anonfun$fetchAndReadPact$4(PactBrokerClient.scala:182) [info] at scala.Option.map(Option.scala:230) [info] at com.itv.scalapactcore.common.PactBrokerClient.fetchAndReadPact(PactBrokerClient.scala:176) [info] at com.itv.scalapactcore.common.PactBrokerClient.$anonfun$fetchPactsFromPactsForVerification$6(PactBrokerClient.scala:61) [info] at scala.Option.map(Option.scala:230) [info] at com.itv.scalapactcore.common.PactBrokerClient.$anonfun$fetchPactsFromPactsForVerification$5(PactBrokerClient.scala:58) [info] at scala.collection.immutable.List.flatMap(List.scala:366) [info] at com.itv.scalapactcore.common.PactBrokerClient.$anonfun$fetchPactsFromPactsForVerification$4(PactBrokerClient.scala:56) [info] at scala.Option.map(Option.scala:230) [info] at

I?ve verified and the mentioned pact file exists and should be found

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

@tjones has left the channel

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

Been trying to figure out the right matching rule for this sitaution. Below is the body ```{ "ok" : true, "systems" : { "GooglePubSub" : { "ok" : "true" }, "Database" : { "ok" : "true" }, "GoogleGroups" : { "ok" : "true" }, "GoogleIam" : { "ok" : "true" }, "OpenDJ" : { "ok" : "true" } } }``` I would want to make sure that each system name (?GooglePubSub? etc) is a string. I know that you can do a matching rule on the ok type, but not really sure how to check the field name

In the case that I can?t check the field name, is there a trick to get a similar effect?

It would probably be a method named something like ?each key like? I think https://docs.pact.io/implementation_guides/jvm/consumer#matching-any-key-in-a-map is what you want. My understanding is that these matchers are exposed to the scale interface yeah?

oh, no

I don?t know if scale ITV does. `pact4s` I think would

If you know the keys in advance, you don?t need to check the types, you can just check that they exist and their values are correct. But I suspect you know this, and that there could be many keys

You could potentially deal with this using Provider states (given that GooglePubSub, Database? exist)

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv

has joined #pact-scala-itv