March 2020 - Haskell-Cafe

stack: pandoc fails to build as dependency
by Olaf Klinke 06 Aug '20

06 Aug '20

Dear cafe, I am baffled to learn that stack does not provide builds as reproducible as advertised. I have a project that depends on pandoc. It builds fine on my Debian 9 machine. Then I check out the project on a Debian 8 machine. Downloading and building pandoc-2.7.3 with the latest stack-2.1.3 on the Debian 8 machine runs fine. Now I create a new project, using the same stack.yaml as in the pandoc-2.7.3 package, except now pandoc-2.7.3 is an extra-dep. The project's only module contains nothing but: module Test where import Text.Pandoc Building this project with the same stack results in a linker error: pandoc > Preprocessing executable 'pandoc' for pandoc-2.7.3.. pandoc > Building executable 'pandoc' for pandoc-2.7.3.. pandoc > [1 of 2] Compiling Main pandoc > [2 of 2] Compiling Paths_pandoc pandoc > Linking .stack-work/dist/x86_64-linux/Cabal-2.4.0.1/build/pandoc/pandoc ... pandoc > collect2: error: ld returned 1 exit status pandoc > `gcc' failed in phase `Linker'. (Exit code: 1) This seems to be independent of the LTS version (the stack.yaml in pandoc-2.7.3 says lts-13.17 although lts-13.17 contains pandoc-2.5 and pandoc-2.7.3 is available only in 14.9). Can somebody hint what I am doing wrong? Should I file a bug report? Thanks Olaf

3 3

Using State monad with lens
by Michael Hull 02 Apr '20

02 Apr '20

Hi, I use the State monad in a lot of places, but when I use it with the lens library, I find myself writing code like: myFunc :: StateT MyState ([ResultType]) myFunc = do st <- get case st ^. some_property of True -> do my_value .= "Hello" True -> do my_value .= "Goodbye" I want to do something based on a 'state-variable', but I have to write 'get' -- is there a pattern that allows me to avoid the intermediate 'st' variable? I want to write something like: myFunc = do case (get ^. some_property) of True -> do my_value .= "Hello" True -> do my_value .= "Goodbye" but that won't compile. Best wishes, Mike

6 6

How to make a Hashable instance for (Finite n)?
by David Banas 31 Mar '20

31 Mar '20

What is the most expedient way to get around this error? Could not deduce (Hashable (Finite n)) Thanks, -db

1 0

Research Assistant/Associate Position (Postdoc) at Imperial College London
by David Castro 30 Mar '20

30 Mar '20

Hello, We are looking for candidates for this position at Imperial College London. The official text of the advertisement follows: Department of Computing, Imperial College London Research Assistant/Associate Position (Full Time) 35,477 GBP to 47,579 GBP per annum Reference: ENG01270 Fixed-term: 2 years (with a possible 24 month extension) Starting date: as soon as possible (no later than 1st October 2020) Closing Date: 26th April 2020 The Research Assistant will work under the EPSRC Established Career Fellowship Project, POST: Protocols, Observabilities and Session Types. Imperial College London provides a flexible arrangement to be able to start and work remotely until we re-open the campus (this depends on your nationality and current living address). Please contact with Nobuko Yoshida (n.yoshida(a)imperial.ac.uk) Imperial College London if you would like to apply to the position to have informal discussions. The application deadline is also extendable. ------------------------------------------------------------ The project has particular emphasis on putting theory into practice, by embedding session types in a range of programming languages and applying them to case studies; or developing the links between session types and other areas of theoretical computer science. The research programme includes collaboration with several companies and organisations. Candidates for the post-doc position will need to have expertise in either: 1. programming language design and implementation; or 2. formal semantics, type theory and concurrency theory Different positions will be suitable for different points on the theory/practice spectrum. We are especially interested in candidates with a combination of theoretical and practical skills. For more details, see http://mrg.doc.ic.ac.uk. The focus of Imperial College London Group is theories and applications of (Multiparty) Session Types (JACM,POPL'08) which include: -- Go (POPL'19,ICSE'18,POPL'17,CC'16), Scala (PLDI'19,ECOOP'17,ECOOP'16), F# (CC'18), Erlang (CC'17), Haskell (CC'20,POPL'16), OCaml (SOC), Java (FASE'16,FASE'17), MPI-C (FPL'16,OOPSLA'15,CC'15) and Python (FOAC,LMCS,FMSD); -- session types theories (ESOP'20,POPL'19,ESOP'19,ESOP'16,CONCUR'15), automata theories (CAV'19,CONCUR'19,FoSSaCs'17,TACAS'16,CONCUR'15,POPL'15), game semantics (POPL'19,FoSSaCs'19), Implicit Complexity (LICS'18) and linear logic (ESOP'18,FoSSaCs'18,CONCUR'15); or -- mechanisation of session types meta-theory (Coq, Isabelle, Agda, etc) (TACAS 20) -- other applications such as blockchains (FSE'19) and robotics (ECOOP'19) The contact person is Professor Nobuko Yoshida, Imperial College London (n.yoshida(a)imperial.ac.uk) The candidate is welcome to contact her. Details: https://www.imperial.ac.uk/jobs/description/ENG01270/research-assistant-res…

1 0

[ANN] esqueleto-3.3.3.0: Support for UNION, safer joins, and subquery joins!
by Christian Charukiewicz 29 Mar '20

29 Mar '20

Hi all, My friend Ben Levy and I spent a while writing and documenting Database.Esqueleto.Experimental, a new module in the esqueleto package which adds support for UNION, subqueries, and improves the safety of outer JOINs. We've done a lot of use/testing of this implementation, but it is currently in the Experimental module in order to allow for gathering feedback. It is now live on Hackage as esqueleto-3.3.3.0. The module docs give examples of the new functionality. Feel free to take a take a look here: https://hackage.haskell.org/package/esqueleto-3.3.3.0/docs/Database-Esquele… Thanks, Christian

1 0

gitlab.haskell.org down?
by Ivan Perez 29 Mar '20

29 Mar '20

Hi, Is the Haskell gitlab down or has anything changed? https://downforeveryoneorjustme.com/gitlab.haskell.org Cheers, Ivan

3 2

Re: [Haskell-cafe] Strange error when cabal installing.
by David Banas 28 Mar '20

28 Mar '20

Okay, I was able to get the _Haskell Platform_ approach to work, by doing the following: - Uninstall Haskell Platform. - Reinstall Haskell Platform, overriding the default suggested installation path with something _not containing spaces_. - Change my `~\AppData\Roaming\cabal\config` file from this: ``` extra-prog-path: C:\HaskellPlatform\8.6.5\msys\usr\bin, C:\Users\capnf\AppData\Roaming\cabal\bin ``` to this: ``` extra-prog-path: C:\HaskellPlatform\8.6.5\msys\usr\bin extra-prog-path: C:\Users\capnf\AppData\Roaming\cabal\bin ``` -db

1 0

Why does the Haskell Platform Windows installer suggest a path with spaces?!
by David Banas 28 Mar '20

28 Mar '20

Why does the Haskell Platform Windows installer suggest a default installation path that contains spaces?! This breaks builds: configure:2365: checking build system type configure:2379: result: x86_64-pc-msys configure:2399: checking host system type configure:2412: result: x86_64-pc-msys configure:2487: checking for gcc configure:2514: result: C:\PROGRA~1\Haskell Platform\8.6.5\lib\../mingw/bin\gcc.exe configure:2743: checking for C compiler version configure:2752: C:\PROGRA~1\Haskell Platform\8.6.5\lib\../mingw/bin\gcc.exe --version >&5 C:\/Users/capnf/Documents/GitHub/gitit/network-3.1.1.1/configure: line 2754: C:\PROGRA~1\Haskell: No such file or directory configure:2763: $? = 127 configure:2752: C:\PROGRA~1\Haskell Platform\8.6.5\lib\../mingw/bin\gcc.exe -v >&5 C:\/Users/capnf/Documents/GitHub/gitit/network-3.1.1.1/configure: line 2754: C:\PROGRA~1\Haskell: No such file or directory -db

3 2

Which option causes 'cabal install' to leave temp. build dirs. in place?
by David Banas 28 Mar '20

28 Mar '20

Hi all, Which option to a 'cabal install' command will cause the temporary build directories to be left in place, so that I can inspect, for instance, the 'config.log' file of a failed dependency build? Thanks, -db

2 1

ANN: latex-svg
by Oleg Grenrus 27 Mar '20

27 Mar '20

I occasionally write blog posts with some math. So far I have been using latex-formulae family of packages. This week Andres Löh asked whether LaTeX can be rendered as SVGs. The answer turns out to be yes, it wasn't few years ago. The preview CTAN package, and the dvisvgm tool make it very easy. I forked latex-formulae into latex-svg to use above mentioned tools. There are three packages on Hackage now: - latex-svg-image: render LaTeX formulae to SVG images - latex-svg-pandoc: render LaTeX formulae in pandoc documents - latex-svg-hakyll: render LaTeX formulae in hakyll pages To mention some differences between latex-formulae and latex-svg: - latex-svg-image uses only latex and dvisvgm, ImageMagick is not required. You don't need to tweak policy.xml to allow rasterization of PostScript. (See this StackOverflow question). - latex-svg-image supports global cache (off by default), which speedups hakyll site builds. My blog builds in 20 seconds instead of three minutes with empty cache. I wouldn't worry about space usage too much, the cache contents take 15M after building my blog. which is way less than the size of site executable. - latex-svg-hakyll has initFormulaCompilerSVGPure variant, which doesn't need IO to be created. You don't need to thread renderFormulae function through. Together with global cache, the perfomance penalty is small. - One drawback is that result pages become bigger (and slower). For example a formula-heavy https://oleg.fi/gists/posts/2018-12-12-find-correct-laws.html is - 576k in size with latex-formulae - 2819k in size with latex-svg (almost 5 times larger). Rendering SVG images is also more CPU expensive. Otherwise the API and module names are the same. If you are already a user of latex-formulae you can easily migrate to latex-svg. See the announcement blog post how some examples look like: https://oleg.fi/gists/posts/2020-03-06-latex-svg.html <http://oleg.fi/gists/posts/2020-03-06-latex-svg.html> <http://oleg.fi/gists/posts/2020-03-06-latex-svg.html> Crisp math-blogging to everyone - Oleg Grenrus <http://oleg.fi/gists/posts/2020-03-06-latex-svg.html>

4 5