lihaoyi/Ammonite

Rock-solid shell scripting

Symhedonic

@Shokodemon

does ammonite still phone home ?

Alexandre Archambault

@alexarchambault

@lihaoyi Just saw lihaoyi/Ammonite#984 (which I guess ought not to slow down things compared to before the addition of --thin)

Just so you know, I tend to run Ammonite via other kinds of coursier bootstraps (without jars-in-jars), lightweight ones that rely on the coursier cache. These don't incur perf penalties, and are actually faster than assemblies (I described how I benchmark things, with benchmark results, in this notebook).

These could be used instead of the assemblies that are pushed to github.

Li Haoyi

@lihaoyi

sounds good, i think as a default just sticking with dumb assemblies might be the best thing for now, just to fix the regression

after the fix, we can then consider fancy additional stuff separately :)

There is some understandability downside going from bog-standard-classloader-isolation, 1995-style, to a new bootstrap/isolation mechanism.

Alexandre Archambault

@alexarchambault

Yeah I agree on that :|

I intend to document them in more detail soon(ish), along with the new install command of coursier.

Li Haoyi

@lihaoyi

when you use Ammonite yourself, e.g. for embedding into almond, that's a sufficiently specialized use case I think it's ok if you end up bundling it yourself (e.g. shading all the non-api classfiles)

Alexandre Archambault

@alexarchambault

I don't shade stuff in almond, I rely on the lightweight coursier bootstraps (that rely on the coursier cache instead of embedded jars in jars).

Li Haoyi

@lihaoyi

:+1:

megri

@megri

@lihaoyi I just used Ammonite + ujson + upickle + requests to rehydrate an elasticsearch index that got destroyed, from unparsed event data stored in a remote cassandra cluster. Took 1 hour all in all, with no previous experience of requests or ujson. I just wanted to say thank you!

I'm already a patreon but I should probably prod my bosses and ask them to add some themselves.

Li Haoyi

@lihaoyi

glad to hear it!

Srepfler Srdan

@schrepfler

:clap:

Olivier Mélois

@Baccata

Has ammonite dropped support for the 2.11 series ? 1.6.9 doesn't seem to have been published for those

(not that I care about having support, but I'm updating mill with ammonite 1.6.9 and some tests are failing because the repl artifact is missing for 2.11)

Li Haoyi

@lihaoyi

yes

I support two major versions :)

feel free to bump whatever failing mill tests to 2.12

Olivier Mélois

@Baccata

Okeydo 👍

Symhedonic

@Shokodemon

trying to parse docker logs, and im getting the following throwbale

java.lang.IllegalArgumentException: Unknown ansi-escape at index 3083 inside string cannot be parsed into an fansi.Str
 fansi.ErrorMode$Throw$.handle(Fansi.scala:417)
  fansi.ErrorMode$Throw$.handle(Fansi.scala:405)
  fansi.Str$.apply(Fansi.scala:270)
  fansi.Str$.implicitApply(Fansi.scala:225)
  pprint.Renderer.$anonfun$rec$27(Renderer.scala:136)
  pprint.Result$.fromString(Result.scala:53)
  pprint.Renderer.rec(Renderer.scala:136)
  pprint.Renderer.$anonfun$rec$15(Renderer.scala:97)
  scala.collection.Iterator$$anon$10.next(Iterator.scala:448)
  pprint.Util$ConcatIterator.check(Util.scala:32)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  pprint.Util$ConcatIterator.check(Util.scala:29)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  pprint.Util$ConcatIterator.check(Util.scala:29)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  pprint.Util$ConcatIterator.check(Util.scala:29)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  pprint.Util$ConcatIterator.check(Util.scala:29)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  pprint.Util$ConcatIterator.check(Util.scala:29)
  pprint.Util$ConcatIterator.hasNext(Util.scala:40)
  scala.collection.Iterator$$anon$12.hasNext(Iterator.scala:500)
  pprint.Truncated.next(Truncated.scala:93)
  pprint.Truncated.next(Truncated.scala:11)
  scala.collection.Iterator$$anon$10.next(Iterator.scala:448)
  scala.collection.Iterator.foreach(Iterator.scala:929)
  scala.collection.Iterator.foreach$(Iterator.scala:929)
  scala.collection.AbstractIterator.foreach(Iterator.scala:1406)
  ammonite.repl.ReplApiImpl.show(ApiImpls.scala:105)
  ammonite.repl.ReplApiImpl.show$(ApiImpls.scala:92)
  ammonite.repl.Repl$$anon$1.show(Repl.scala:76)
  ammonite.repl.ReplApiImpl.show(ApiImpls.scala:88)
  ammonite.repl.ReplApiImpl.show$(ApiImpls.scala:88)
  ammonite.repl.Repl$$anon$1.show(Repl.scala:76)
  ammonite.$sess.cmd125$.<init>(cmd125.sc:1)
  ammonite.$sess.cmd125$.<clinit>(cmd125.sc)

is there any way i could preprocess the strings and get rid of those characters?

Symhedonic

@Shokodemon

i realized i can circumvent the errors using lines.map(line => (printf("s", line))

Li Haoyi

@lihaoyi

easiest would be to just println them yourself

you could also customize repl.pprinter to not try to be clever with strings, but that's more work (not that much more tho)

Andrew Richards

@ajrnz

Does the ammonite/mill's import $ivysyntax have anything equivalent to sbts classpathTypes which lets you import extra types of artifacts (just just jar/source etc)?
I am unable to use:

import $ivy.`org.vafer:jdeb:1.7`

because it declares its artifact to be type maven-plugin (not jar) and is ignored by coursier.

In sbt you can fix this with:

classpathTypes += "maven-plugin"

Any workaround?

Olivier Mélois

@Baccata

@ajrnz if you manage to download the artifact "manually" (there might be an option in the coursier API, otherwise you'll have to build the http request by hand), then you can call upon the interp api to load stuff into the classpath. That artifact is then taken into account for the compilation of the next stage of your script (or the downstream scripts) : http://ammonite.io/#Multi-stageScripts

Andrew Richards

@ajrnz

@Baccata Yeah that works. Not pretty though:

coursier fetch org.vafer:jdeb:1.7

and then

import $ivy.`org.vafer:jdeb:1.7`
interp.load.cp(Seq(os.Path(coursier.cache.CacheDefaults.location + "/https/repo1.maven.org/maven2/org/vafer/jdeb/1.7/jdeb-1.7.jar")))
@

It not obvious how/where you would insert this into the ammonite code

Vladyslav Pekker

@agilesteel

hi I just downloaded ammonite for the first time and am trying out the source command which doesn't seem to work at all

@ source(Some(1).map(identity)) 
Unable to find sourcecode for class scala.Option

is this a known issue? I'm on Ubuntu 18.04 in case that matters

Alexandros Bantis

@ambantis

Hi, I'm running ammonite 1.6.6 (2.11.12). How do I have the ammonite script exit with a non-zero exit code?

Olivier Mélois

@Baccata

@ambantis you can always call System.exit(1)

Symhedonic

@Shokodemon

hello all

i would like to use repl.sess.save and repl.sess.load in my scala code, - they are visible from the predef, but i would like to use them inside a class that belongs to a different package.

from what i understand, they become available after the shell is started with Main(), but id like the following piece of code to compile as well"

//  def getArtificats(
//                                                        service: String,
//                                                        serviceVersion: String,
//                                                        serviceDependencyVersion: String,
//                                                        namespace: String
//                                                      ) {
//
//
//    val sessionTag = s"$service:$serviceVersion:$serviceDependencyVersion"
//    repl.sess.save(sessionTag)
//
//    val dep = coursier.
//      Dependency(
//        coursier.Module(namespace, service),
//        serviceVersion,
//        attributes = coursier.Attributes(classifier = "fat")
//      )
//
//    printf(s"finished fetching dependencies for $service-$serviceVersion with dependencies from version $serviceDependencyVersion")
//    repl.sess.load(sessionTag)
//    printf(s"session $sessionTag restored ")
//  }

Symhedonic

@Shokodemon

not found: value repl
[error]     repl.sess.save(sessionTag)
[error]     ^

i am using ammonite 1.3.3 and importing as a dependency in SBT, and initializing it via Main(). .

Christopher Davenport

@ChristopherDavenport

Is there a trick to run this with 2.13 presently that I’m missing?

Li Haoyi

@lihaoyi

no? the default instructions are for 2.13

it’s on maven central for 2.13

no tricks required

Yann Moisan

@YannMoisan

hello. I have some .sc files. What do you suggest to unit test code in these files ?

aappddeevv

@aappddeevv

Is there a way to have ammonite automatically reload from a classpath changed files? I was looking through the docs but did not see anything that mentions this. My scenarios is similar to that using jupyter. I typically need to develop a library but need to test and play with the data interactively while building it out--a bit like TDD but more exploratory. So there is a separate continuous build task running on the library code but I am working at the repl. Python has something like this where it re-evaluates all imported packages prior to evaluating a user-entered expression.

Dan Di Spaltro

@dispalt

is there a 2.12.9 version available?

Binh Nguyen

@ngbinh

I remember there is a way to use ammonite version that was published for scala 2.12.8 with build that set scala version to 2.12.9 but I do not recall now

Amr Hassan

@superamr_twitter

Hello. Can we get a release for 2.12.9 please? :)

Victor Ivri

@vivri

Hi there, I'm using Ammonite 1.6.9 and Scala 2.13. For some reason, Futures seem to hang, as well as the (ported) parallel collections. I saw old posts about it; has this been fixed? If so, how?

Paul Snively

@paul-snively

Does anyone know how I would go about embedding Ammonite in an app, loading a .sc file from a directory, and ensuring definitions in the .sc are visible (by name) to the app? In other words, a classic “dynamic linking” scenario.

It looks like some combination of instantiateInterpreter and processModule heads in the right direction, but the details seem hazy, especially around accessing bindings in the .sc.

Li Haoyi

@lihaoyi

@PaulAtFormation I don’t think there’s any builtin way to do that, you have to poke around the internal-ish code. Look at how Mill does it (RunScript.scala or something like that)

basically you evaluate the script and get an untyped reference to the wrapper object. Mill then uses reflection to call stuff on the wrapper object, but you could just as easily cast the wrapper object to an interface (you’ll need to make it extend the interface via a custom CodeWrapper) and then call methods directly

Where communities thrive

People

Repo info

Activity