- Haskell-Cafe - Haskell.org

Hyper-recursion?
by Lawrence Bottorff 17 Jan '17

17 Jan '17

A while back I worked at an assessor's office, i.e., the people responsible for handling properties as land parcels. I was brought in to do GIS. One of the things we wanted was to get a sense of the history of land use, specifically how property changed hands, and especially how property lines changed due to properties either merging or being split up. That is to say, how the parcel map changed over time. Problematic was how the rugged old mainframe system (IBM AS/400 and a Cobol app) simply updated the data -- with no concept of remembering how things used to be. My GIS system (ESRI) also lacked any state memory -- other than simply snapshooting everything ad nauseum. Years later I become familiar with the world of functional programming. Somewhere in the functional paradigm, specifically recursion, would seem to be a model for this issue. That is, the whole issue of parcels changing through merges or splits might be a simple question of recursion. If functional is "beyond state," then changes in a map would only be stages, places in a never-ending outward-growing recursion. Seen as "infinite mirrors," one map change would simply be a mirror layer -- click on it to call it up. So in chapter 1 of any functional programming tutorial is the factorial calculation function done with recursion. We beginners see the recursion "going out" to the "last one," then "coming back," adding up the results of each stage as it returns . . . like a yo-yo winding out, then winding up again. But what I'm describing with changing parcels maps would seem to be a process that never comes out of recursion, rather, lives inside a "hyper-recursion." That is to say, an overarching recursion is the whole world. However, today's mentality would simply say, *Obviously, you want to save off each change you make to your parcel map *. . . again, the whole snapshot idea. This thinking is typical data management coupled, perhaps, with the human penchant to see things in terms of time and timelines. So, again, the standard solution would be to timestamp a parcel change and save it off, i.e., a "change" is a tick on the timeline, mark it, make note of it. But again, couldn't we see this whole issue of parcel changes as (using my yo-you analogy) just levels in what seems like a continual winding out -- which in a topological sense doesn't even need to be seen as geographical? This would seem to put the *process in a recursion,* rather than *recursion(s) in a process.* Haskell being the most pure functional language, is, hence, my starting point on this question. I'm a bright enough person, but I can't believe I'm the first person to notice this Alice in Wonderland situation. . . BTW, we could see a chess game in a similar light. . . . if not many, many other evolving phenomena of supposed (cumulative, consequential?) change. Thanks, Lawrence Bottorff

3 2

BayHac 2017 announcement
by Dan Burton 15 Jan '17

15 Jan '17

I am pleased to announce that BayHac 2017 will be held at the Takt office in San Francisco on the weekend of Friday, April 7th thru Sunday, April 9th. BayHac is a Haskell hackathon weekend event for the San Francisco Bay Area and Silicon Valley. It will be a weekend of learning, hacking, and connecting with fellow Haskell enthusiasts. See the page for BayHac 2017 on the haskell.org wiki for details: https://wiki.haskell.org/BayHac2017 -- Dan Burton

2 1

LambdaConf 2017 CFP
by John A. De Goes 15 Jan '17

15 Jan '17

The LambdaConf 2017 Call for Proposals <http://surveymonkey.com/r/lambdaconf-2017-cfp> (CFP) is open until the end of January: http://surveymonkey.com/r/lambdaconf-2017-cfp LambdaConf is an interdisciplinary functional programming conference held in the Mountain West, and seeks to bring together aspiring and practicing functional programmers, as well as researchers. Traditionally, the conference has had a very large amount of Haskell content (second only to Scala content in some years). The conference takes place May 25 - 27, in Boulder, Colorado, and is preceded by commercial training opportunities and a day of third-party mini-conferences on selected topics. If you are an educator, a researcher, a speaker, a speaker coach, or someone aspiring to one of the preceding, then we warmly welcome you to submit a proposal for LambdaConf 2017. All LambdaConf speakers receive free registration, an appreciation dinner, a special gift, accommodations for up to 4 nights (May 24 - May 28), and ample travel assistance based on need and availability. LambdaConf attracts everyone from the FP-curious to researchers; hobbyists and professionals; academics and commercial developers. Material at all levels, including beginner content and very advanced content, will find an audience at LambdaConf. LambdaConf welcomes the following kinds of proposals: Languages. Proposals that overview or dive into specific features of functional, math, or logic programming languages (both new and existing), with the goal of exposing developers to new ideas or helping them master features of languages they already know. Libraries. Proposals that discuss libraries that leverage functional or logic programming to help programmers solve real-world problems. Concepts. Proposals that discuss functional programming idioms, patterns, or abstractions; or concepts from mathematics, logic, and computer science, all directed at helping developers write software that's easier to test, easier to reason about, and easier to change safely. Aspects. Proposals that discuss how functional programming can help with specific aspects of modern software development, including scalability, distributed systems, concurrency, data processing, security, performance, correctness, user-interfaces, machine learning, and big data. Use Cases. Proposals that discuss how functional programming enabled a project or team to thrive, or deliver more business value than possible with other approaches. Cherry-Picking. Proposals that show how techniques and approaches from functional programming can be adapted and incorporated into mainstream development languages and practices, to the benefit of developers using them. Cracks. Proposals that call attention to difficulties of functional programming (both as a cautionary tale but also to raise awareness), especially such proposals that suggest alternatives or a path forward. Data. Proposals that present data, measurements, or analysis that suggests different techniques, paradigms, languages, libraries, concepts, or approaches have different efficacies for given specified metrics, which provide actionable takeaways to practicing functional and logic programmers. Off-Topic. Proposals that have appeal to a mainstream developer audience (the number of off-topic proposals we accept is small, but we do accept some, especially for keynotes). LambdaConf welcomes proposals for dynamic and static programming languages, category theoretic and otherwise, both functional and logic, or mainstream languages that can benefit from functional programming. Challengers and alternatives are also welcome. We are currently soliciting for the following session formats: Speaker Coaches. Speaker Coaches have extensive experience preparing material and giving polished, well-structured, educational, entertaining talks. Speaker Coaches review material and practice sessions for 5 - 10 speakers, & provide constructive feedback to speakers about how to improve the focus, content, & presentation of material. Coaches receive accommodations, free registration, & airfare reimbursement, & a 10 minute speaking slot (if interested). Leap Workshops. Leap Workshops take place on May 25, and are approximately 6 hours in length. They are in-depth, hands-on workshops designed to teach mainstream functional programming topics in enough detail, attendees can immediately apply what they learn in their jobs. Instructors for Leap Workshops receive accommodations, free registration, airfare reimbursement (including family), and a small stipend. If you are an author with a relevant book, we can also make it easy for attendees to purchase the book as part of registration for your workshop. Workshops must use a containerized cloud environment provided and paid for by LambdaConf. Hop Workshops. Hop Workshops are 2 hours in length. Like Leap Workshops, these workshops are in-depth and hands-on, but they cover reduced content and may be specialized to topics that may not have mainstream appeal. Guides receive accommodations, free registration, and travel assistance based on need and availability. Workshops must use a containerized cloud environment provided and paid for by LambdaConf. De Novo Sessions. De Novo Sessions are 1 hour in length. These sessions are designed to present original work from industry and academia. While the requirements for proposals are more rigorous, there is less competition for De Novo slots. Speakers receive accommodations, free registration, and travel assistance based on need and availability. Educational Sessions. Educational Sessions are 1 hour in length. These sessions are designed to clearly and concisely teach one useful concept, skill, aspect, library, or language to attendees. Teachers receive accommodations, free registration, and travel assistance based on need and availability. Inspire Talks. Inspire talks are 10 minutes in length and focus on clear communication of a single takeaway. These sessions are intended to inspire attendees to learn more about particular subjects or to try new approaches. Speakers receive accommodations, free registration, and travel assistance based on need and availability. Note: While only 10 minutes in length, we require a high-level of polish, structure, and rehearsal for Inspire Talks; despite shorter length, they are as much work to prepare for as other sessions. Keynotes. Keynotes are 40 minutes in length, and are presented before all attendees (there are no other sessions concurrent with keynotes). Keynotes are designed to offer thought-provoking, opinionated, and insightful commentary on topics of interest to the community. Keynotes have more latitude to stray from functional programming and into related areas such as computer science, mathematics, and science. Keynote speakers receive accommodations, free registration, and travel assistance based on need and availability. LambdaConf requires tailored content. This does not mean you cannot reuse existing content, only that you should incorporate updates into your material, respond to feedback from your speaker coach, and tailor the material to the LambdaConf audience as appropriate. Note: The Call for Proposals closes January 31. Submitters will be notified by March 1st at latest. -- John A. De Goes john(a)degoes.net Follow me on Twitter @jdegoes

1 0

The design and use of QuickCheck
by Joe Nelson 15 Jan '17

15 Jan '17

Hey Haskellers, I decided to learn more about QuickCheck because it seems useful, and after a two weeks of reading papers and trying experiments, I wrote about my findings. https://begriffs.com/posts/2017-01-14-design-use-quickcheck.html Hope you find the article useful. Cheers, Joe

1 0

Re: [Haskell-cafe] Strange encoding issue with Text and Servant
by David Turner 14 Jan '17

14 Jan '17

Ok, so there's definitely a difference of some kind client-side. It'd be useful to see the network traffic nonetheless, as the source of the '!D(MISSING)' is still a bit of a mystery and you may see something suggestive in the raw bytes. I can't find the string 'MISSING' anywhere in servant or its dependencies, even looking in libs and binaries. Any chance of tracking its source down using `find` and `grep` and `strings`? On 14 Jan 2017 19:48, "Jan von Löwenstein" <jan.loewenstein(a)gmail.com> wrote: Actually the code I pointed at _is_ the swagger-codegen generated client. I have got a very small wrapper around it: ``` listService :: (MonadIO m, MonadCatch m, MonadReader Config m, MonadLog Text m) => Text -> m ServiceList.ServiceList listService labelSelector = namespacedF $ \namespace -> Kube.listNamespacedService namespace Nothing (Just labelSelector) Nothing Nothing Nothing Nothing listSecret :: (MonadIO m, MonadThrow m, MonadReader Config m, MonadLog Text m) => Text -> m SecretList.SecretList listSecret labelSelector = namespacedF $ \namespace -> Kube.listNamespacedSecret namespace Nothing (Just labelSelector) Nothing Nothing Nothing Nothing namespacedF :: (MonadIO m, MonadThrow m, MonadReader Config m, MonadLog Text m) => NamespacedF model -> m model namespacedF f = do config <- ask result <- let baseUrl = apiEndpoint config tlsSettings = Base.tlsSettings baseUrl (credentials config) kubeNamespace = namespace config in do manager <- liftIO $ newManager $ (mkManagerSettings tlsSettings Nothing) { managerModifyRequest = \req -> return req { checkResponse = \req res -> do Text.IO.hPutStrLn stderr (Text.pack (show req)) responseMessage <- showResponse res Text.IO.hPutStrLn stderr responseMessage return () } } liftIO $ runExceptT $ f kubeNamespace manager baseUrl either throwM return result ``` I call them like: ``` do let Just agentId = ... selector = "agentId" <> "=" <> agentId logMessage $ "listSecret selector: '" <> selector <> "'" secretList <- listSecret selector ``` `logMessage` gives a literal equals sign in both cases. `Text.IO.hPutStrLn stderr (Text.pack (show req))` from namespacedF gives `%3D` and `%!D(MISSING)` depending on which function I call. >From what I understand this is definitely happening on the client side. Do you still think a tcpdump might be worth it? ``` listSecret selector: 'agentId=b49d4406-8020-44e6-7788-6f92d5c0e732' Request { host = "192.168.99.100" port = 8443 secure = True requestHeaders = [("Accept","application/json")] path = "/api/v1/namespaces/default/secrets" queryString = "?labelSelector=agentId%!D( MISSING)b49d4406-8020-44e6-7788-6f92d5c0e732" method = "GET" proxy = Nothing rawBody = False redirectCount = 10 responseTimeout = ResponseTimeoutDefault requestVersion = HTTP/1.1 } ``` ``` listService selector: 'agentId=24f99a4b-1682-44da-7ba4-dba935d107d2' Request { host = "192.168.99.100" port = 8443 secure = True requestHeaders = [("Accept","application/json")] path = "/api/v1/namespaces/default/services" queryString = "?labelSelector=agentId%3D24f99a4b-1682-44da-7ba4- dba935d107d2" method = "GET" proxy = Nothing rawBody = False redirectCount = 10 responseTimeout = ResponseTimeoutDefault requestVersion = HTTP/1.1 } ``` Thanks for the time you already invested. I am completely puzzled because I see no chance how the code could behave as it does. (From my experience that points out I am looking at the wrong place ;) ) Best Jan David Turner <dct25-561bs(a)mythic-beasts.com> schrieb am Sa., 14. Jan. 2017 um 20:08 Uhr: > Grasping at straws a little bit here, but can you (a) do a packet capture > (e.g. `tcpdump -X`) to see what's going back and forth on the wire, just to > make absolutely sure all of the oddness is on the client's end? > > A string like `!D(MISSING)` sorta looks like a decoding error so perhaps > it will be instructive to look at the bytes on the wire it's trying to > decode, although I can't think what could be confused with an equals sign. > It's not like that blooming Greek question mark ( > http://www.fileformat.info/info/unicode/char/037e/index.htm) > > Cheers, > > > > > On 14 January 2017 at 18:47, Nigel Rantor <wiggly(a)wiggly.org> wrote: > > On 14/01/17 17:37, Jan von Löwenstein wrote: > > Oh, it happens with servant on the client side. > > I use a swagger-codegen generated servant based client to access > kubernetes. > Calling https://github.com/soundcloud/haskell-kubernetes/blob/ > master/lib/Kubernetes/Api/ApivApi.hs#L475 works > Calling https://github.com/soundcloud/haskell-kubernetes/blob/ > master/lib/Kubernetes/Api/ApivApi.hs#L473 doesn't. > > From what I see the `QueryParam"labelSelector" Text`is the same in both. > > Logging the value gives the literal `=` as expected. Logging the > (http-client) requests shows the difference. > > > So I went ahead and wrote a simple client by hand and when I call the two > functions on a local endpoint that shows me the request URL I get the > following. > > Same arguments for 'pretty' and 'labelSelector' in both cases. > > Code: > > runListSecret :: Manager -> BaseUrl -> ClientM SecretList > > runListSecret manager baseUrl = listSecret (Just "prettyArg") (Just "=") > Nothing Nothing Nothing Nothing manager baseUrl > > > > runListService :: Manager -> BaseUrl -> ClientM ServiceList > > runListService manager baseUrl = listService (Just "prettyArg") (Just "=") > Nothing Nothing Nothing Nothing manager baseUrl > > Server log: > > 127.0.0.1 - - [14/Jan/2017:18:39:49 +0000] "GET /api/v1/secrets?pretty=prettyArg&labelSelector=%3D > HTTP/1.1" 404 - 0.0003 > 127.0.0.1 - - [14/Jan/2017:18:39:49 +0000] "GET /api/v1/services?pretty=prettyArg&labelSelector=%3D > HTTP/1.1" 404 - 0.0003 > > I haven't used swagger-codegen but the servant-client code works for me so > I would look more into that. > > Swagger codegen looks like it might take me a while to get my head into so > maybe you could look at the generated code, or sling it to me here or > directly. > > The two calls are so similar that I can't see why the generated client > would behave differently in those cases. > > Regards, > > n > > Best > Jan > > Nigel Rantor <wiggly(a)wiggly.org <mailto:wiggly@wiggly.org>> schrieb am > > Sa., 14. Jan. 2017 um 18:11 Uhr: > > On 13/01/17 19:28, Jan von Löwenstein wrote: > > Hi, > > > > I have got a two places with a `QueryParam "q" Text` and call it > with a > > Text that contains a `=` literal. In one place the `=` is correctly > > encoded as %3D, in the other I see a `!D(MISSING)`. > > > > This has to happen somewhere in Servant or the lower layers. Both > Texts > > print out nicely with an `=` sign if I just print them to stdout. > > > > Google does not find `!D(MISSING)` anywhere. > > > > Any idea what could possibly be the problem here? > > It would be really useful to see exactly how you are 'calling' this. > > Do you have a curl command line or are you using some other client? > > n > > _______________________________________________ > Haskell-Cafe mailing list > To (un)subscribe, modify options or view archives go to: > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe > Only members subscribed via the mailman list are allowed to post. > > > _______________________________________________ > Haskell-Cafe mailing list > To (un)subscribe, modify options or view archives go to: > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe > Only members subscribed via the mailman list are allowed to post. > > >

1 0

Calling Haskell from C
by Dominic Steinitz 14 Jan '17

14 Jan '17

I've been trying to get the example in https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/ffi-chap.ht… working. I have a cabal file name: test-via-c version: 0.1.0.0 homepage: TBD license: MIT author: Dominic Steinitz maintainer: idontgetoutmuch(a)gmail.com category: System build-type: Simple cabal-version: >=1.10 executable Foo.dylib main-is: Foo.hs other-extensions: ForeignFunctionInterface build-depends: base >=4.7 && <4.10, primitive >=0.6 && <0.7 hs-source-dirs: src default-language: Haskell2010 include-dirs: src ghc-options: -O2 -shared -fPIC -dynamic extra-libraries: HSrts-ghc8.0.1 which installs /Users/dom/Library/Haskell/ghc-8.0.1/lib/test-via-c-0.1.0.0/bin/Foo.dylib I then compile the C with gcc-6 Bar.c /Users/dom/Library/Haskell/ghc-8.0.1/lib/test-via-c-0.1.0.0/bin/Foo.dylib -I/Library/Frameworks/GHC.framework/Versions/8.0.1-x86_64/usr/lib/ghc-8.0.1/include -L/Library/Frameworks/GHC.framework/Versions/8.0.1-x86_64/usr/lib/ghc-8.0.1/rts -lHSrts-ghc8.0.1 Sadly when I try to run the executable I get ~/Dropbox/Private/TestU01-1.2.3/examples/src $ ./a.out dyld: Library not loaded: @rpath/Foo.dylib Referenced from: /Users/dom/Dropbox/Private/TestU01-1.2.3/examples/src/./a.out Reason: image not found Trace/BPT trap: 5 I am guessing the executable is looking for the dynamic library. I copied Foo.dylib to the current directory but still get the same error. I know this is really an OS X question but I am hoping someone who has done before will be able to help. Dominic Steinitz dominic(a)steinitz.org http://idontgetoutmuch.wordpress.com

4 8

module is a package module error
by Lana Black 14 Jan '17

14 Jan '17

Hi, I'm trying to use hint [1] to compile in runtime externally supplied haskell code. The compiled code must have certain modules imported. The problem is that I'm getting this error when loading modules >WontCompile [GhcError {errMsg = "<no location info>: error: module >\8216Internal\8217 is a package module"}] I'm not really sure what to do with this. One way to fix this is to move Internal to a separate package so that hint could load it through cabal, but that's something I'd like to avoid. Is there any way I could make hint use the same modules from the package it's being called from? The code is located at [2]. [1]: http://hackage.haskell.org/package/hint [2]: https://github.com/greydot/hint-test/

2 3

Feedback needed: CLI app + guide
by Carlo Matteo Scalzo 13 Jan '17

13 Jan '17

Hi all, I've already posted this in beginners@, but I thought it might be of interest to a wider audience too (apologies for double-posting). I started learning Haskell about a year ago, and I couldn't find any example of a simple, real world application written in Haskell, so as usual I ended up writing my own :-) The code is available for free (it's a simple command-line app), and I've also written a short guide that uses the application as a case study to talk about Haskell's features. Could you guys please download the guide (for free, of course) and let me know if you find it useful? You can get the guide here: https://fifohaskell.com Please send me your feedback either here or to my personal email address (it's on the website), and thank you very much for your help! Carlo

1 0

brainstorm help
by Michael Litchard 13 Jan '17

13 Jan '17

Hi There. I'm looking for ideas to flesh out a writing project I'm beginning. What I am looking for is a set of five games in increasing complexity that would allow for a wide discussion of Haskell. I originally began this project by centering the writing around one large game, but this idea has proven to be brittle. The set of five should include a discussion of: (1) Types (2) folds, maps, filters (3) monoids, functors, applicatives, and monads Thanks for what ever storms your brains bring forth. :)

7 11

Can I allow multiple rigid type variables?
by Cody Goodman 13 Jan '17

13 Jan '17

I know my question doesn't quite make sense and you can't have "multiple rigid type variables" but I couldn't think of a better way to characterize it or explain it. So I'll also try to give more detail below along with the code example and error at the end. I have an inner join function using vinyl which does cool stuff like giving a type error if you try to join on a column that doesn't exist. I previously had an implementation that blindly built an index based on the left argument. I then tried to dispatch based on the lengths of the left and right producer arguments. The problem is my return type is: Rec f (as ++ bs) That means I have two different producers, one of as and one of bs. I then do a case statement on the lengths and I want to store the producer to use for a given piece in that variable. The problem is ghc "pins" the type variable as either (Rec f as) or (Rec f bs) when I need to make that choice at runtime. My code and error is both at github here: https://github.com/codygman/frames-diff/blob/inner-join-fix/Frames/Diff.hs#… and below: -- | Performs an inner join and keeps any duplicate column -- Recommend keeping columns in producers disjoint because accessing -- anything but the leftmost duplicate column could prove difficult. -- see: https://github.com/VinylRecords/Vinyl/issues/55#issuecomment-269891633 innerJoin :: (MonadIO m, Ord k) => Producer (Rec f leftRows) IO () -- leftProducer -> Getting k (Rec f leftRows) k -- leftProducer lens -> Producer (Rec f rightRows) IO () -- rightProducer -> Getting k (Rec f rightRows) k -- rightProducer lens -> m (P.Proxy P.X () () (Rec f (leftRows ++ rightRows)) IO ()) innerJoin leftProducer leftLens rightProducer rightLens = do leftProducerLen <- P.liftIO $ P.length leftProducer rightProducerLen <- P.liftIO $ P.length rightProducer let curProducer = case rightProducerLen < leftProducerLen of True -> rightProducer -- False -> leftProducer let curKeymapProducer = case rightProducerLen < leftProducerLen of True -> leftProducer -- False -> rightProducer let curLensLookup = case rightProducerLen < leftProducerLen of True -> rightLens -- False -> leftLens let curLensInsert = case rightProducerLen < leftProducerLen of True -> leftLens -- False -> rightLens let appender km row = case rightProducerLen < leftProducerLen of True -> case M.lookup (view curLensLookup row) km of Just otherRow -> pure $ rappend otherRow row Nothing -> P.mzero -- False -> case M.lookup (view curLensLookup row) km of -- Just otherRow -> pure $ rappend row otherRow -- Nothing -> P.mzero keyMap <- P.liftIO $ P.fold (\m r -> M.insert (view curLensInsert r) r m) M.empty id curKeymapProducer pure $ curProducer >-> P.mapM (\r -> appender keyMap r) -- error if I uncomment my false cases (specifically the False case for curProducer) -- [5 of 5] Compiling Frames.Diff ( Frames/Diff.hs, interpreted ) -- Frames/Diff.hs:125:32: error: -- • Couldn't match type ‘leftRows’ with ‘rightRows’ -- ‘leftRows’ is a rigid type variable bound by -- the type signature for: -- innerJoin :: forall (m :: * -> *) k (f :: * -- -> *) (leftRows :: [*]) (rightRows :: [*]). -- (MonadIO m, Ord k) => -- Producer (Rec f leftRows) IO () -- -> Getting k (Rec f leftRows) k -- -> Producer (Rec f rightRows) IO () -- -> Getting k (Rec f rightRows) k -- -> m (P.Proxy P.X () () (Rec f (leftRows ++ rightRows)) IO ()) -- at Frames/Diff.hs:113:14 -- ‘rightRows’ is a rigid type variable bound by -- the type signature for: -- innerJoin :: forall (m :: * -> *) k (f :: * -- -> *) (leftRows :: [*]) (rightRows :: [*]). -- (MonadIO m, Ord k) => -- Producer (Rec f leftRows) IO () -- -> Getting k (Rec f leftRows) k -- -> Producer (Rec f rightRows) IO () -- -> Getting k (Rec f rightRows) k -- -> m (P.Proxy P.X () () (Rec f (leftRows ++ rightRows)) IO ()) -- at Frames/Diff.hs:113:14 -- Expected type: Producer (Rec f rightRows) IO () -- Actual type: Producer (Rec f leftRows) IO () -- • In the expression: leftProducer -- In a case alternative: False -> leftProducer -- In the expression: -- case rightProducerLen < leftProducerLen of { -- True -> rightProducer -- False -> leftProducer } -- • Relevant bindings include -- curProducer :: Producer (Rec f rightRows) IO () -- (bound at Frames/Diff.hs:123:7) -- rightLens :: Getting k (Rec f rightRows) k -- (bound at Frames/Diff.hs:119:47) -- rightProducer :: Producer (Rec f rightRows) IO () -- (bound at Frames/Diff.hs:119:33) -- leftLens :: Getting k (Rec f leftRows) k -- (bound at Frames/Diff.hs:119:24) -- leftProducer :: Producer (Rec f leftRows) IO () -- (bound at Frames/Diff.hs:119:11) -- innerJoin :: Producer (Rec f leftRows) IO () -- -> Getting k (Rec f leftRows) k -- -> Producer (Rec f rightRows) IO () -- -> Getting k (Rec f rightRows) k -- -> m (P.Proxy P.X () () (Rec f (leftRows ++ rightRows)) IO ()) -- (bound at Frames/Diff.hs:119:1)

2 3