has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

Though I guess this really may as well be in the standalone group

Just an update on where this is at. The code required in the standalone to generate and verify a message pact is out - however, the v3 pact spec is only partially supported. You won't be able to verify any jvm message pacts yet.

At the moment, the parser is expecting one provider state string instead of an array of objects.

:the_horns:

I'm not far off having beta's of both JS and Go to test the waters. I have a few questions about the implementation that we can discuss at that point here

Does it properly support matching yet Beth?

I noticed this when trying matchers: ``` { "consumer": { "name": "MyConsumer" }, "provider": { "name": "MyProvider" }, "messages": [ { "description": "a dog", "providerState": "some state", "content": { "id": { "contents": 1, "json_class": "Pact::SomethingLike" }, "name": { "contents": "rover", "json_class": "Pact::SomethingLike" }, "type": { "data": { "generate": "bulldog", "matcher": { "json_class": "Regexp", "o": 0, "s": "bulldog|sheepdog" } }, "json_class": "Pact::Term" } } } ], "metadata": { "pactSpecification": { "version": "2.0.0" } } } ```

But, the verifier properly caught the matcher

Should, but just the v 2 ones.

Hm. The spec version isn't right

I figured at least that, but was expecting to see the matching rules in there

ah

Are you passing it in to the command line?

that?s from Pact node. Pretty sure that?s in other stuff too :slightly_smiling_face:

There's a bug then.

It must be missing a connection.

ok will take a look

bug in JS/Node or Ruby?

ah, be back soon?

Ruby I think, because the specification version is 2.0.0 in the file

Ah

Confirmed. I can see where it is.

Will fix when I get a chance

Cool, thanks

No hurries.

It should still verify, yes?

Also, we'll need to pass in 3.0.0 when writing the message pacts.

It does, albeit o need to test the regex matcher as it wasn't quite behaving is I expected. That could he me so I need to confirm a few thibgs

Will do re 3.0.0

Hmm, I was just looking - there is currently no option to pass in a spec version to `pact-message` - where should it be passed in?

There is a command line option

ah, so there is

I was invoking the help command incorrectly

Thanks for that `reify` command Beth, works a charm

Cool

I?ll update Pact JS/Node when I get a moment, but that?s working so won?t prioritise it

There are a couple of things we should discuss, which I?ll summarise in here shortly for the other implementations

For some reason though, the `--pact-dir` argument doesn?t like certain paths. e.g. this fails: ``` pact-message update '{"content":{"access":{"contents":{"role":{"data":{"generate":"admin","matcher":{"json_class":"Regexp","o":0,"s":"admin|controller|user"}},"json_class":"Pact::Term"}},"json_class":"Pact::ArrayLike","min":3},"id":{"contents":127,"json_class":"Pact::SomethingLike"},"name":"Baz"},"providerState":"some state","metadata":{"Content-Type":"application/json; charset=utf-8"},"description":"some test case"}' --consumer billy --provider bobby --pact-dir /Users/mfellows/go/src/github.com/pact-foundation/pact-go/examples/pacts/ /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-message-0.4.0/lib/pact/message/consumer/interaction_decorator.rb:36:in `extract_content': undefined method `content' for #<Pact::Interaction:0x007ff211a27000> (NoMethodError) from /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-message-0.4.0/lib/pact/message/consumer/interaction_decorator.rb:18:in `as_json' from /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-message-0.4.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `block in as_json' from /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-message-0.4.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `collect' from /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-message-0.4.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `as_json' from /opt/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.6.4/lib/pact/consumer_contract/consumer_contract_decorator.rb:28:in `to_json' from /opt/pact/lib/vendor/ruby/2.2.0/gems/json-2.1.0/lib/json/common.rb:286:in `generate' from /opt/pact/lib/vendor/ruby/2.2.0/gems/json-2.1.0/lib/json/common.rb:286:in `pretty_generate' ``` but this is totes fine: ``` pact-message update '{"content":{"access":{"contents":{"role":{"data":{"generate":"admin","matcher":{"json_class":"Regexp","o":0,"s":"admin|controller|user"}},"json_class":"Pact::Term"}},"json_class":"Pact::ArrayLike","min":3},"id":{"contents":127,"json_class":"Pact::SomethingLike"},"name":"Baz"},"providerState":"some state","metadata":{"Content-Type":"application/json; charset=utf-8"},"description":"some test case"}' --consumer billy --provider bobby --pact-dir /tmp/pacts ```

ahh

Was there an existing doc there?

ya

It may have been trying to merge

I think it must have been an old format?

Yes, not compatible.

@bethskurrie OK, so my pact-message is failing after the last update: ``` Diff -------------------------------------- Key: - is expected + is actual Matching keys and values are not shown { - "id": 1, - "name": "rover" + "id": 27, + "name": "fido" } Description of differences -------------------------------------- * Expected 1 but got 27 at $.id * Expected "rover" but got "fido" at $.name ``` My pact file seems to have the correct v3 matchers, however: ``` { "consumer": { "name": "MyJSMessageConsumer" }, "provider": { "name": "MyJSMessageProvider" }, "messages": [ { "description": "a request for a dog", "providerStates": [ { "name": "some state" } ], "content": { "id": 1, "name": "rover", "type": "bulldog" }, "matchingRules": { "body": { "$.id": { "matchers": [ { "match": "type" } ] }, "$.name": { "matchers": [ { "match": "type" } ] }, "$.type": { "matchers": [ { "match": "regex", "regex": "^(bulldog|sheepdog)$" } ] } } } } ], "metadata": { "pactSpecification": { "version": "3.0.0" } } } ```

Will have a look now

ah, I know. I?m probably not specifying the version to the verifier, perhaps?

You definitely need that

But it may or may not be the cause

No, the verifier should read the version from the pact

ah nah that doesn?t make sense - there isn?t an option for it. exactly

Let me look. It's probably me.

thanks

I've got the same problem. It's not recognising the matching rules. Will investigate.

kk

good to know not just me!

I was also going to ask Beth, given the format looks very v3 speccy, does this mean we have interoperability with Pact JVM generated messages or do we need to do more work on that front?

The missing part (once I've implemented the metadata) will be the multiple provider states, and the provider state params. It will *read* a v3 file from pact jvm and not blow up, but it will ignore the bits it can't understand.

cool, that?s a great compromise and starting point. Probably worth a `WARN` in the logs so people can see, and we can update client side docu about v3 compatibility

Yup, it's prefixed with WARN

`WARN: Using only the first provider state, "an alligator named Mary exists", as support for multiple provider states is not yet implemented. Ignoring provider states: "another state", "yet another state"`

`WARN: Ignoring params for the following provider states as params support is not yet implemented: "yet another state"`

To the error stream, so it won't screw up JSON parsing.

:thumbsup:

amaze

Hopefully it won't be in there for long, but as you say, all code that is released to production, stays in production!

Rust is dead, long live Ruby :wink:

Ha

The rust impl seems to have lost momentum

I think Ron almost has full v3 spec implemented, but we?ll be fighting inertia and all of the other niceties we have with Ruby

@bethskurrie how far off are we for making the Message Stuff _real_ for Ruby devs? Is it good to go? Does it just need docs? Or does the DSL etc. need a facelift?

It just needs a DSL.

ok cool

I've done the underlying logic, obviously, but there isn't a nice interface for it, esp on the producer side.

got it

Someone was asking about it weren't they?

someone else did too (can?t remember) but the team at Stichfix are mostly Ruby, and they?re _keen as mustard_

Ok, I'll get on to it.

I was mostly not bothering because I didn't think anybody cared

:cry:

People care - have you not seen the almost 2k stars on the repo :smile:

Mm.. mostly for historical purposes though.

It was the original http://docs.pact.io really.

true

The knee high dictators have arisen. I must hearken to their commands.

(there doesn?t appear to be a :salute? but salute)

has joined #pact-message-ruby

has joined #pact-message-ruby

@matt.fellows, I am spending sometime catching up to the JS implementation and the spikes with @bethskurrie on message support. What is the thought on adding properties to the pact message objects like Message.ProviderStates when those are not in the original pact-specification v3 (https://github.com/pact-foundation/pact-specification/blob/version-3/testcases/message/body/matches.json).

@bethskurrie, is there a /message endpoint on the ruby standalone? When the framework registers the interaction with the ruby standalone, how will it know to turn a pact-v3 with a message node, rather than an interactions node?

As another thing I am working through, I am on Windows ( :violin: ) with Pact Ruby Standalone for 1.47.1 and submitting the JSON from the demo: https://github.com/pact-foundation/pact-message-demo/blob/master/message-pact.json ... and I get an error:`{ "message": "Error ocurred in mock service: NoMethodError - undefined method `[]' for nil:NilClass", "backtrace": [ "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-support-1.6.2/lib/pact/consumer_contract/interaction.rb:32:in `parse_v2_interaction'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-support-1.6.2/lib/pact/consumer_contract/interaction.rb:26:in `from_hash'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/mock_service/request_handlers/interaction_post.rb:24:in `respond'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/mock_service/request_handlers/base_request_handler.rb:17:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/rack-2.0.5/lib/rack/cascade.rb:33:in `block in call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/rack-2.0.5/lib/rack/cascade.rb:24:in `each'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/rack-2.0.5/lib/rack/cascade.rb:24:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/consumer/mock_service/cors_origin_header_middleware.rb:11:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/consumer/mock_service/error_handler.rb:13:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/mock_service/app.rb:33:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/pact-mock_service-2.9.0/lib/pact/consumer/mock_service/set_location.rb:14:in `call'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/rack-2.0.5/lib/rack/handler/webrick.rb:86:in `service'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/webrick-1.3.1/lib/webrick/httpserver.rb:138:in `service'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/webrick-1.3.1/lib/webrick/httpserver.rb:94:in `run'", "C:/Users/me/Downloads/pact-1.47.1-win32/pact/lib/vendor/ruby/2.2.0/gems/webrick-1.3.1/lib/webrick/server.rb:191:in `block in start_thread'" ] }

There is no mock for message pact.

The developer defines the message (and matchers) then calls a language appropriate thing (eg "yield_message") to tell the pact code to give it the concrete message, and then the test makes sure that the prod code does the right thing with the message.

Sorry I didn?t see this. I think perhaps that just means it?s missing from the spec tests (cc @uglyog) because they are definitely in the implementation

e.g. https://github.com/pact-foundation/pact-specification/tree/version-3#allow-multiple-provider-states-with-parameters

has joined #pact-message-ruby

The current test cases only test the matching rules

any reason?

because that is all we wrote :smile:

I assumed as much

Thanks, y'all

I am attempting at understanding the Message Pact Validation on https://github.com/pact-foundation/pact-message-demo and implement in Pact-PHP. Let me see if I can understand the control flow and character Pact Provider Verifier. This is pact-provider-verifier(.bat on Windows) in the ruby stand alone. Correct? This will be called by Pact-JS, http://Pact.Net, and Pact-PHP through a message specific implementation. Producer Proxy HTTP Server. This is created by the writer of the test using the Pact libraries (i.e. Pact-JS, http://Pact.Net, and Pact-PHP). Message Invoker. This is a call to the part of the Producer functions Pact libraries (i.e. Pact-JS, http://Pact.Net, and Pact-PHP) that is likely a call back or reference to the Producer class and function that produces appropriate message. Producer class is the function under test. My question is ... is the Producer Proxy HTTP Server written by the implementer of the test or the Pact-(PHP|JS|Net) creator? It would seem really odd for a implementer of the test to be forced to write an HTTP server if their app. Thus, I assume Producer Proxy HTTP Server is actually built into Pact-PHP (in this case). Am I understanding the ownership correctly?

It's written by the framework authors (i.e. us)

We tell the provider verifier to read a given message pact, and to send the message object to the proxy. All we really use from this is the description field (the provider state flow is similar). The proxy then takes that message name and invokes the specific message producer that inplements that message. The response is extracted and sent back to the verifier who then checks it matches whats in the pact spec.

Sorry @cfmack the proxy http server is written by you

I thought so (and actually hoped so) as that would be _terrible_ for the library users.

:smile:

yeah, pretty harsh!

We thought of a number of ways we could do this. If you have better ideas we?re open to it

One of the main benefits of this approach is it isn?t message specific, so it means other protocols and transport even can be introduced this way

The http proxy is there to enable a cross language process interop call - it's not ideal, and will be much more elegant when we can do it with the rust bindings.

but, like the whole ruby standalone thing, it will get the job done!

The message invoker is the part that the test writer will need to implement (either via a DSL or just a plain old <insert language here> class)

I'm there! Thanks, y'all. Got any ideas on the rest of the v3 spec? Are we implementing that soon or should we just call this spec "2.1"?

It's a good point. I've left it at beta for now but if we foresee a long delay perhaps that's best

To end my day, the implementation of creating the proxy server in a scripting language blows my mind. What does the url and payload look like? We are spinning up a server in a library to dynamically load classes with a dynamic function to accept a dynamic payload that is likely complicated by directory locations of the thing calling the library, which likely has dynamic includes. :exploding_head:. No way simple way to make this secure other than putting a big ol' asterisk on it. I'm going to have to noodle on this one

What's the security angle I'm missing?

> What does the url and payload look like? The URL can be whatever you like, as you're going to be configuring both that and the call to the provider verifier. The payload is ``` { description: "description", providerStates: [{ name: "name", params: {} }] } ```

Hey @matt.fellows, I have a POC complete where I have a listener setup, spins up a proxy provider server in the background, kicks off the unit test, runs pact-provider-verifier.bat, calls the proxy provider server with a POST per message. I am exploring how to tell the proxy server what class to call (i.e. the call back). It looks like in Pact-JS, we set this up as configuration vs a per request. Do you spin up proxy provider per test? I would love to spin up one proxy server then reuse it based on the call back. However, it does not look like I could get the callback information in there via the pact-provider-verifier.bat call

NVM, l am going to explore using a customer-provider-header

That works. Ignore that question. Now for my next reflection ...

pact-provider-verifier.bat returns the description and providerState of each message. A common simple case across message queues would be: Consumer: I need a message that looks like {id : 1} on the queue/routing_key/message_key called "my_key" Producer: I publish "my_key" Hey, let's write a pact and agree on that! It would seem natural for the Consumer and Producer to put the information about "my_key" in the metadata and {id: N} in the contents. Right? If my throught process plays out, pact-provider-verifier.bat would need to somehow pass over the metadata, not just description and providerState. Neither description or providerState seem natural to put "my_key". I suppose you would argue in some cases that is a providerState but a queue seems to be pretty straightforward. Would a description be "A queue with my_key exists" and providerState is "my_key"? If that holds true, what is metadata for?

... and the customer-provider-header does not work if a pact has more than one message with different call back functions

What is the customer provider header?

Tbh, I never understood the metadata. I think @uglyog uses it for a content type.

Whats the deal with the queue names Ron?

Is that a part of the contract?

Some message queues allow for metadata to be applied to the message (ie. routing info)

and the content type comes from there

I have people asking for the metadata to be verified as well

I'll have to add that then.

Currently, I'm only verifying the content.

Should queue names be included somewhere?

it could go into the metadata

But I don't see the value

@cfmack I tend to see the http proxy as something that lasts the entire length of the pact verification process. Starting and stopping it would slow things unnecessarily I think.

I think _customer_ -> _custom_

The Golang proxy tends to spin up in < 10ms, so it?s not something I?m super concerned with. That being said, I use the same proxy instance throughout the process.

Oh, right. What problem are you trying to solve with the custom provider header @cfmack? Is it something we can provide proper support for?

Stepping back, queue name parallels the same concept at a URI path. "I expect your API to have an end point at /foo" "I expect to listen to your messages on queue 'foo'"

I have to find away to pass a call back process to an outside process. I was going to pass in that information via the custom provider header. That falls apart when the callback needs to change. I'll try to establish a better use case.

So if we add matching to the metadata, you can put it there.

If you get a repo that demonstrates the problem, I can probably find a way to support you.

The fact that matt's done the js and go impls without a callback suggests to me that there is a way without it, however.

That works. My next question is ... how does the provider validate when it is only passed the data description & providerState? Should it need to know about the metadata? Maybe not. Let me think about that one. Should that just be in the description and validated on the metadata? (I like to work these things out as I type)

The only external process should be the verifer itself

The proxy should wrap any test code in process, so no need for dynamic callbacks etc

:smile:

The description and state should tell you what message to return I think.

Right now I took the decision _not_ to validate the metadata, if only because I don?t think it?s well defined in the spec just yet and I couldn?t see a nice way to pass it to the function

The matching is performed by the provider verifier process.

Sorry, on the consumer side this is

Oh, right.

IFF i created protocol specific handlers (e.g. AMQP, gRPC or something) that needed it, I would have to ensure it did the right thing.

Presumably the verifier _would_ match it?

It could, but currently does not.

I didn't really understand it.

I can easily add verification for it. But we'd want it to be in line with @uglyog code

I have an outstanding feature request to add it to the jvm

I think I got it y'all. I have a functioning Provider with limitations. I'll say that Go and Js are not the same for async programming as PHP, especially on Windows. That said, I got an initial stab on it and have passing tests. Thanks for your help.

I wonder if we have a maintainers hangout and do a demo of each language so we can all see how each works. This will help us get consistency across experiences for our users. Thoughts?

Probably easier somehow

If you want to torture your eyes, here's the crux of the implementation on the provider side: https://github.com/pact-foundation/pact-php/blob/feat/message-pact/example/tests/MessageProvider/MessageProviderTest.php https://github.com/pact-foundation/pact-php/blob/feat/message-pact/src/PhpPact/Provider/MessageVerifier.php

After getting this slowly up to snuff (lots to do still), I was thinking about getting a v2 implementation in R. Why? Because it would be fun. Anyone interested?

Really good idea @matt.fellows

@cfmack imo porting one of the current wrapper impls to the rust impl is a more productive use of time, but if you're doing it for fun, then who am I to object!

I pick fun things over necessary things more then I care to admit.

We all do. What's life without whimsy

has joined #pact-message-ruby

@neil if you want to do some Pact Message stuff, check out the following two links: * https://gist.github.com/bethesque/c858e5c15649ae525ef0cc5264b8477c * https://github.com/pact-foundation/pact-message-demo

There should be enough context to get an idea of how it works, albeit potentially @cfmack has feedback on the use-fullness of them (as he was the most recent beneficiary of this)

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

@bethskurrie @matt.fellows I have a request to have the message metadata to be able to be verified along with the contents. I have picked this issue up now (https://github.com/DiUS/pact-jvm/issues/479). This will need to be implemented in the Ruby implementation.

Sounds good. Is it just the ability to pass the metadata along to the provider or does the verification do some extra processing?

Can do @uglyog

Matchers and generators can also be applied to the metadata

How does the verification of the metadata work, though? Can you show some pseudocode of what you?re thinking/doing?

Umm, yeah, about that. When I get to that part I'll let you know :grin:

has joined #pact-message-ruby

@matt.fellows just finished the verification of message metadata

Example consumer test: https://github.com/DiUS/pact-jvm/blob/master/pact-jvm-consumer-junit5/src/test/java/au/com/dius/pact/consumer/junit5/AsyncMessageTest.java#L31

Corresponding pact file JSON

``` { "description": "a test message", "metaData": { "destination": "X001", "Content-Type": "application/json; charset\u003dUTF-8" }, "contents": { "testParam1": "value1", "testParam2": "value2" }, "providerStates": [ { "name": "SomeProviderState" } ], "matchingRules": { "metadata": { "destination": { "matchers": [ { "match": "regex", "regex": "\\w+\\d+" } ], "combine": "AND" } } } } ```

The invoked test methods need to now return both the message contents and actual metadata

For backwards compatibility, if only the contents are returned, the metadata is ignored and not verified

Example test code https://github.com/uglyog/pact-gradle-test/blob/master/src/test/groovy/com/github/uglyog/pactgradletest/GroovyMessageTest.groovy#L15

@uglyog noted

has joined #pact-message-ruby

On the pact message stuff, I was just talking to @murphy.meng who was asking about the request/response async pattern

My understanding is that to achieve this currently, you need to write two separate tests - one for the request flow and another for the response flow (where the consumer and provider take turns initiating the request)

See recent comments https://gist.github.com/bethesque/c858e5c15649ae525ef0cc5264b8477c for others sharing this sentiment

yes I did read what Alex commented there, ?it will require two message pacts and four time-separated commits for each request-response pair.? It means what you just said about two separate tests? I don?t quite understand how this work around work at the moment.

I mean, I might want to try out pact and two sepaate tests are fine. Could you explain a bit in detail?

I?m not sure about the ?time-separated commits? bit

you would need to create one test for `message from A` -> `[ queue ] ` -> `received by B`, and then write a separate test for the response pathway `B sends response` -> `[queue]` -> `received by A`

If we had the concept of a request/response message here, you would just need to write one test for this scenario

I am going to ask another dumb question, you said pact-mock_service is not used in message api testing, then for example in `message from A` -> `[ queue ]` -> `received by B` scenario, where is the mocked receiver (to replace real B) and generate the pact for A?

The framework will still perform this action, it just may not use the mock service - does that matter to you?

It does not matter your are right it is just impl detail. I simply want to understand better:wink:, sorry. when you say framework, you mean ruby impl, go, javascript and jms impl? Do they share the same impl and they do their own impl?

haha np

In this case, I mean whatever language bindings you are using. So in the case of Pact JS, the JS library will actually invoke your consumer and then invoke another Ruby process asking it to serialise the Pact file. The mock service is not involved

(`pact-message` binary does the serialisation here)

Cool. I believe if I add message impl to Python I will need know how to invoke pact-message binary but let?s put it aside for now.

My next dumb question is, which ruby binary will be used to verify the pact with messaging api scenario?

right, so are you wanting to contribute? This would have been helpful information!!

@murphy.meng start here

I am very keen to allow Python to support messaging api, I managed to get some bandwidth to do this for testing the messaging api in my company.

that?s great

I didn?t realise that was your intention

See the sequence diagrams at https://github.com/pact-foundation/pact-message-demo for background

Sure will do.

has joined #pact-message-ruby

I'm running into a problem when testing messages with pact-net, and the output seems to suggest the problem originates in the ruby core. I've got a test that succeeds when running from Windows but fails when running from Linux because it can't parse json. This is a consumer test so I guess the json is generated during the test and not input *for* the test. Does this sound familiar at all? I don't really know Ruby myself so I'm kind of stuck here. ```/mnt/c/Repos/Pact-Net/Samples/MessageZooEvents/ZooEventsConsumer.Tests/bin/Debug/netcoreapp2.2/pact-linux-x86_64/lib/vendor/ruby/2.2.0/gems/json-2.1.0/lib/json/common.rb:156:in 'parse': 757: unexpected token at ''{id:{json_class:Pact::SomethingLike,contents:2},name:{json_class:Pact::SomethingLike,contents:Terence},type:{json_class:Pact::SomethingLike,contents:Giraffe}}'' (JSON::ParserError) from /mnt/c/Repos/Pact-Net/Samples/MessageZooEvents/ZooEventsConsumer.Tests/bin/Debug/netcoreapp2.2/pact-linux-x86_64/lib/vendor/ruby/2.2.0/gems/json-2.1.0/lib/json/common.rb:156:in 'parse' from /mnt/c/Repos/Pact-Net/Samples/MessageZooEvents/ZooEventsConsumer.Tests/bin/Debug/netcoreapp2.2/pact-linux-x86_64/lib/vendor/ruby/2.2.0/gems/json-2.1.0/lib/json/common.rb:335:in 'load' from /mnt/c/Repos/Pact-Net/Samples/MessageZooEvents/ZooEventsConsumer.Tests/bin/Debug/netcoreapp2.2/pact-linux-x86_64/lib/vendor/ruby/2.2.0/gems/pact-message-0.5.0/lib/pact/message/cli.rb:24:in 'reify'``` https://github.com/pact-foundation/pact-net/pull/175

It looks like it's trying to parse an object that is already parsed.

Why this should happen on Linux and not Windows I don't know!

@bart.schotten can you raise this in the pact-ruby-standalone repo with the full backtrace, and I'll have a look at it.

Thanks for the quick response! Here it is: https://github.com/pact-foundation/pact-ruby-standalone/issues/32

I'll try and get to it as soon as I can. Things are a bit busy here with the launch of http://pactflow.io so forgive me if it's not immediately.

No problem, I appreciate the work you do :+1:

You're welcome.

I recall having an issue with reify early on in the implementation - i?ll check my notes tonight to see if anything turns up

From memory, I got it when I gave it an invalid JSON payload (I think maybe we were iterating on the property keys)

It looks like it's lost its quotes

yeah, looks to have double

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

Hi guys/gals, are there examples on how to integrate pact for messages for ruby

@khoatran now I've refreshed my memory, I've remembered the DSL for message pact is not done in Ruby yet. The underlying logic is done though, because it's being used by pact-js. Which side would you be interested in using? Consumer or provider, or both?

hi Beth, I'm interested in using the provider side

we have some ruby code that pushes messages to a queue, and some node.js code consuming the messages on the other end

It shouldn't be too much work to hook up the verification code.

I'll try to get on to it in the next week or so.

Thank you Beth

@khoatran sorry, I've been caught up with other things. I'll have a look at it tomorrow.

@khoatran If you update to the latest pact gem you'll have pact message verification support

Thanks @bethskurrie

@khoatran you'll find the docs here https://github.com/pact-foundation/pact-message-ruby#provider

And here's a running example: https://github.com/pact-foundation/pact-ruby/blob/master/spec/support/message_spec_helper.rb

Sorry @bethskurrie I've been away. Thank you for your help :thumbsup:

has joined #pact-message-ruby

Hey guys, added a small PR to pass through message metadata: https://github.com/pact-foundation/pact-message-ruby/pull/12 - let me know if you have any concerns.

has joined #pact-message-ruby

has joined #pact-message-ruby

Hi @dan.garland the code for the consumer side of the message pact is actually done, but not released as I got into naming conundrums.

There ended up being a "message builder" on both consumer and provider sides, but they did different things and I couldn't work out what to call them, so I didn't release the code :joy:

Srsly, naming things is the hardest.

What you can do to help it get out is work with me on the naming!

I'll send you the code when I have a moment.

@dan.garland https://github.com/pact-foundation/pact-message-ruby/blob/master/spec/features/create_message_pact_spec.rb

I'm trying to think of another name for the `builder` here https://github.com/pact-foundation/pact-message-ruby/blob/master/spec/features/create_message_pact_spec.rb#L11

And I'm really not sure about the syntax for supplying the message pact to the test

This blog outlines how the groovy DSL works for pact-jvm https://dius.com.au/2017/09/22/contract-testing-serverless-and-asynchronous-applications/

And this is how it's done in JS https://github.com/pact-foundation/pact-js/blob/master/examples/messages/consumer/message-consumer.spec.ts

Thanks I tried making a consumer test that followed this and I'll have a think about the points :point_up: . When I looked at the resulting Pact I noticed that the `providerStates` and `matchingRules` were empty- is this the WIP part?

Hm. It might just be unfinished.

I know the underlying impl works because pact-js uses it - it's just been the ruby DSL that was missing.

Yup, there's a TODO in there! https://github.com/pact-foundation/pact-message-ruby/blob/master/lib/pact/message/consumer/interaction_builder.rb#L26

Interesting I'll take a look

In the meantime, I've been thinking more about how Pact would fit in with messaging. It seems that the Pact approach of knowing who your producer/provider and consumers are and naming them in the Pact makes total sense with REST, but perhaps a lesser extent with messages. If you buy into the eventing model, part of the sell is that you can throw messages onto the event bus and not care about other services and how they respond. In this situation the contract is a data contract based on the message content, rather than being between specific providers/consumers. I wondered whether anyone using this feature on other languages had a view on this and whether the Pact approach maps well to messaging in their experience

You raise a good point. The current eventing model misses two key interaction types: req/res style and also the "stream" style you've just described.

I think it's worth revisiting both, but we need to initially get parity with current libraries.

has joined #pact-message-ruby

has joined #pact-message-ruby

Hello, I am trying to use the `pact-message` gem but it seems I'm hitting a wall, due to a weird `method_missing` error. Any tip would be appreciated. I have created a POC repo to reproduce the issue: https://github.com/ang3lkar/poc-kafka-message-contracts/

Hi @angelos if you read the docs, you'll see that the consumer dsl is not finished

hi @bethskurrie! Thanks for answering! However, I'm a bit confused, there is no consumer-side code in my poc, it's just the provider verification of a static contract file. Essentially, it's the simplest manifestation of the provider example in the pact-message README. (https://github.com/pact-foundation/pact-message-ruby/blob/master/README.md#provider)

Is there something I am missing?

has joined #pact-message-ruby

has joined #pact-message-ruby

any news about that?

Hey guys! Awesome job that you guys are doing here! is the idea to move this gem inside `pact-ruby` after finishing it? also: I'm receiving an error while trying to use the message provider: ```undefined method `honours_pact_with' for main:Object```

work around: define a `service_provider` block with your `honours_pact_with` method and then define a `message_provider` block with the `builder` method. Traced the error back to https://github.com/pact-foundation/pact-ruby/blob/master/lib/pact/provider/configuration/message_provider_dsl.rb#L24

https://github.com/pact-foundation/pact-ruby/issues/215 issue opened :slightly_smiling_face:

@max.forasteiro I was planning on keeping them separate

Thanks again for your help on the DSL btw.

has joined #pact-message-ruby

has joined #pact-message-ruby

Hi All, I m trying to do consumer message contract testing but it failing with below error: MessageHandler get_message allows a consumer to test that it can handle a message example correctly (FAILED - 1) Failures: 1) MessageHandler get_message allows a consumer to test that it can handle a message example correctly Failure/Error: message_provider_kafka .given("Encrypted Msg") .is_expected_to_send("payload for file download") .with_metadata("retry_count": 0) .with_content(message: {:headers=>{:message_class=>"notification-newfile", :message_type=>"notifications", :account=>{:id=>"dhgdjhgf"}, :object_id=>"a05f3b28-9056-493d-9ae6-eb67d04ed511", :transaction_id=>"a61d0c8f-ee3f-49fd-9371-ba7d9e86c419", :encryption_key=>"HpHNuKx0GFRSLH6i7NGxHbSqKAOCX-sBtP3pmG1eWHE22WhYeEb9HSlc8RpgTvvJkvmjpsad9AHcd6M9BjI_jHsDv3zmTPL807O99S4PodMRBcxxXlh_LQLKzaXCMCZiZ-nmw-dH5BlY8nmtnAE4VqmOHKMnwU_lQhyXBLb4R4aMqBhjrFZchw9YPDvoO1x8wQIGCXhMp0dnZRWnqHoXR1Pzv9OfEwKIMispJ-wblNAUOEq6UYWbhtr-81jvBBlIjIqiLZUWaOns3fK6xIKLXUQ-TqtfALH7zvUrbXqjeAsfXZZs49Swzthg2yyU8cQvXVMj_NJ6i8mVnCkZ62AmT5dkz9bMOEPWVrnENgxc-4cJjgmfRb2-Bf7vVNqLPis5bSNLtwOTehKbJA-UPjm23wI5pWiuNaHVnbG97S46I0eVt3h4m12a5-3BsAsWT9-YB3X57jh5lCkPz8Z8GsvXGCqrIenw0_W0qJFaiAxjUtmiNJi_Y2kjxyLAejL8YesWWY8wLm24FRUC00Vu0sC8DC-PgQEJmHjyNVeIiAVFh2IOq5fTtbr0k_mAnE_BdzMBoYvDIrXLdu1Hq_FDAcbkoQAxRECRWPemi3wFRhry7Cm75pDolKZcv18n210TVfjrlhCXIvySlQv1FqP3PygQ3FHReX58pTMvXrhEyM9p-jI=", :key_version=>1}, :metrics=>{:created=>"2020-07-03T09:47:28.427570739Z", :encoded=>"0001-01-01T00:00:00Z", :decoded=>"0001-01-01T00:00:00Z", :entry=>"2020-07-03T09:47:28.468636405Z", :exit=>"0001-01-01T00:00:00Z"}, :payload=>"NGB4LZ9A76SEeXEPwv--jwtM6421UeR29JZwbTGW9juYSZYsrRkyyxVxzuumZvFpWnSLNLvfMF9ZWhf8ZyOIT2ksc0muPi1AnicJpDez9WM5qvaua2HAs0FRnrGP1ZxmVJq2j7v9NWgtfG9MnaiebTCH7O6CubJSmYnr0y-kO9V1--DVI64tVDIZgrRax8CE-up5CAlOS4yvkX6CKVQPd6BaDdCynFRFbdyO9F6-tiQfJFOakztTLkh_FQQhL-25egtPf9TdByEY"}) NoMethodError: undefined method `contents' for #<Pact::Interaction:0x005640992bfe28> # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-message-0.7.0/lib/pact/message/consumer/interaction_decorator.rb:37:in `extract_contents' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-message-0.7.0/lib/pact/message/consumer/interaction_decorator.rb:18:in `as_json' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-message-0.7.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `block in as_json' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-message-0.7.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `collect' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-message-0.7.0/lib/pact/message/consumer/consumer_contract_decorator.rb:14:in `as_json' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-mock_service-3.6.1/lib/pact/consumer_contract/consumer_contract_decorator.rb:28:in `to_json' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/json-2.3.1/lib/json/common.rb:521:in `generate' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/json-2.3.1/lib/json/common.rb:521:in `pretty_generate' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-mock_service-3.6.1/lib/pact/consumer_contract/consumer_contract_writer.rb:76:in `pact_json' # /home/vitthal/.rvm/gems/ruby-2.4.0/gems/pact-mock_service-3.6.1/lib/pact/consumer_contract/consumer_contract_writer.rb:66:in `block in update_pactfile'

Consumer is not implemented yet @nirupma.rim

I had hoped to get it out recently, but this whole virus thing has taken away the dedicated oss day in my usual schedule.

okay Thank you

How did you get the syntax if you didn't see the notice in the docs?

I checked the inner implementation

Gosh, not many people ever think to do that!

@bethskurrie: any ETA on completion or what work needs to be done to finish it

If you poke through the consumer code, you'll find a particular dsl/builder class that has mostly empty methods.

I'm sorry, I had hoped to get it out weeks ago. My OSS contribution time used to be on my train trips to and from work, and now I don't have that time.

The coronavirus has no respect for my productivity!

has joined #pact-message-ruby

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-message-ruby

@bethskurrie@bethskurrie@bethskurrie If an intrepid pair of coworkers wanted to pick up working on the ruby kafka consumer what are some links you might suggest reading first? In particular: anything we should do to ensure we follow your lead stylistically?

Hi @armorsmith42, sorry, I only just saw your message.

I'm trying to remember what's outstanding on this. I was going to rename the builder object on the consumer side to avoid confusion with the builder on the provider side.

I'm sure there was some functionality missing too. But not much. I'd have to look through the code again.

If you were able to sketch that out, I would be grateful.

I'll have a look today

I picked it up recently, but my oss time is 0 at the moment, with the kids at home due to lockdown.

Have not forgotten, but didn't get to it today. On the list for tomorrow

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

@bethskurrie@bethskurrie@bethskurrie would you be willing to ?pair program? with me to draft the docs for this?

@armorsmith42 yes, thanks!

If you want to open a PR we can work on it together. It should be fairly straightforward to reverse engineer from the example spec.

has joined #pact-message-ruby

:wave: hi folks, I?m experimenting with the use of pact for contract testing for a kafka consumer and producer. Very stoked to see this gem available, but I am finding it rough going getting things set up (probably me being relatively new to pact in general, and completely new to using it for message-driven systems). Is my understanding correct that in principle it should be possible to write a consumer rspec with a mocked message provider using this gem? Or is it still a work in progress? I found the code example mentioned in the README but I am having a hard time translating it to my situation.

Yeah, sorry, the docs are a bit thin @jeen.broekstra

You don't mock out the message provider, you just write a contract for the message handler

ah! Thank you, that sounds very useful - I hadn?t found that particular page yet :slightly_smiling_face: :book:

there are a lot of docs!

most of them are very http focussed, as that's what came first.

I?m still struggling a little with this. Sorry to flood your channel but let me share a quick consumer-side pact spec that I?m trying to get working. I?ve tried several varieties but what I currently have is this: ```require "rails_helper" RSpec.describe MyConsumer, pact: true do include Pact::Helpers before(:all) do Pact.message_consumer "My Consumer" do has_pact_with "My Provider" do mock_provider :my_provider do pact_specification_version "2" end end end end let(:projector) { double } subject { MyConsumer.new(projector: projector) } describe "processing message", pact: :message do it "does the thing" do my_provider .given("there is a new employee named John").is_expected_to_send("a new employee message") my_provider.send_message_hash { |message| subject.process(message) } expect(projector).to have_received(:handler_for) end end end``` When I try to run this (with `bundle exec rspec`) it gives me an error on the first line of the `it` body: ``` NameError: undefined local variable or method `my_provider' for #<RSpec::ExampleGroups:MyConsumer::ProcessingMessage:0x00007fbb7e647058>``` I?m probably overlooking something trivial but I could use a pointer :slightly_smiling_face:

Try putting your Pact.message_consumer somewhere else

the convention is to put it in a file called "spec/pact_helper.rb"

or, even just put it at the top of the file (to test it)

it doesn?t seem to help unfortunately: ```require "rails_helper" require "pact/consumer/rspec" require "pact/helpers" Pact.message_consumer "My Consumer" do has_pact_with "My Provider" do mock_provider :my_provider do pact_specification_version "2" end end end RSpec.describe MyModule::MyConsumer, pact: true do include Pact::Helpers let(:projector) { double } subject { MyModule::MyConsumer.new(projector: projector) } describe "processing message", pact: :message do it "does the thing" do my_provider .given("there is a new employee named John").is_expected_to_send("a new employee message") my_provider.send_message_hash { |message| subject.process(message) } expect(projector).to have_received(:handler_for) end end end``` gives the same error (also if I put the message consumer in spec/message_providers/pact_helper.rb). Some of the documentation mentioned that I should be working with a builder for the message, rather than a mock provider, but I got a bit confused on how that would work at the consumer side. Am I mixing things up here?

the builder is on the provider side

Looking at this, I can only see that there's a missing require

`require "pact/message/consumer/rspec"`

that might have the pact: :message mixin

:facepalm:

that was it! thanks so much :blush:

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

has joined #pact-message-ruby

:blobwave: Are you using pact-message? Would you like to collab on work to introduce the pact ffi (rust core) into pact-message-ruby and the other Pact gems? I?ve already published a couple of gems with the pact_ffi shared lib being loaded into the namespace with ruby?s `ffi` gem. https://github.com/YOU54F/pact-ruby-ffi I?ve learned Ruby by example from our codebases and happy to pair,

has joined #pact-message-ruby

has joined #pact-message-ruby