@MDXF Does Cubically support its code page or does it not?

@Dennis could you pull Broccoli? Thanks

@ASCII-only Done.

@ASCII-only Definitely didn't break argparsing?

@Pavel >_< IDK, checking now

Seems to work :P

@Pavel I'm assuming flags are eaten by NDesk?

@Pavel ok so what hacky stuff should I do

@Pavel yay

@ASCII-only NDesk doesn't modify the input array. It returns an array of unprocessed arguments.

@Pavel :| then i think i broke something

@ASCII-only Right. It proccess -c twice and doesn't return it. But: tio.run/##S04szclMy8kvTy36/1@joCgzr0TBwTHIPUzz////urr/ddPy8//…

@Pavel ok so what do you mean by modules

@Marco Markdown doesn't work on multiline messages

wow that's an insane commit rate

@Dennis Could you add Egel? According to the README you just need to <dir>/build.sh (once) and run <dir>/src/egel

@Dennis no, you need to install a couple of things but since you're on Fedora just run install.sh.

@Dennis Java 10 is out. You can get the oracle version from oracle.com/technetwork/java/javase/downloads/…

Although if you want to redirect Java 9 to Java 10 and still have OpenJDK, you'd have to build it from source from the OpenJDK repos

@Marco Could you make the prefix of your install script configurable? Installing to /usr is not an option for me.

OJDK 10 repo

@Marco Also, you shouldn't be installing anything to /usr/bin or /usr/lib from not the package manager, use /usr/local instead.

@Marco Is there an equivalent of EGEL_INCLUDE for the library path? I can make it find eg.prelude, but not io.ego.

Copying them to the current directory seems to be a workaround.

@Marco tio.run/#egel

@Pavel How backwards-compatible is Java? Can I run source code intended for Java 9 in Java 10 without problems?

@Dennis Extremely backwards compatible. Three probably haven't been breaking changes in like a decade at least.

I vaguely recall something about _ no longer being an identifier.

Oh, maybe.

The only possibly breaking change in Java 10 is that var became an illegal type name

Java 10 is already stable, yes?

But convention says types must be uppercase anyway

@Dennis Oracle JDK is. OpenJDK is final release candidates.

@Dennis can you pull TIS?

@Dennis stackoverflow.com/questions/16506411/…

@Phlarx Done.

@Pavel OK, so replacing 9 with 10 shouldn't be a big deal.

Found binaries here: jdk.java.net/10

@Dennis Thank you

Is OpenJDK called JDK now?

Or is that something else?

@Dennis jdk.java.net is Oracle JDK

Openjdk is still openjdk.java.net

Ah.

Practical difference?

@Dennis There are a few really minor differences, mostly in internal APIs that aren't part of the Java specification and shouldn't be used.

So could I replace OpenJDK 9 with JDK 10?

Yeah

There was some point when JDK and OJDK had completely different font rendering APIs but I think that got fixed and wouldn't matter on TIO anyway.

How is it possible that there is no jdk.java.net/10?

@Dennis https vs http

Yes, how is it possible that they do not offer HTTPS?

Oh. ¯\_(ツ)_/¯

HTTPS download links on an HTTP site are pointless.

@Dennis This has https and is the same thing: oracle.com/technetwork/java/javase/downloads/…

Also has an rpm

And you have to click a radio button to make the download link available, which appears to be dynamic.

I'm thinking about java-openjdk9 with java-jdk and make it always the latest version. java-openjdk would remain whatever system package Fedora offers.

That makes sense

Do I have to install libraries of will JDK 10 use those of OpenJDK 8?

And assuming I can't find an automated way to download the RPM, how do I make Java 10 find /opt/java-jdk/lib?

@Dennis iirc Java on TIO doesn't know about libraries anyway, and you have to manually give compiler flags specifying their path. That would still work for Java 10.

@Dennis I think you can set the CLASSPATH environment variable

@Pavel Only if you know where the stuff's at.

@Pavel Does CLASSPATH work recursively? I vaguely remember having issues with that.

@Dennis That's still true for Java 8

But it probably shouldn't be. I don't know anything about Java. What doesn't work but should?

@Dennis tio.run/… would need a compiler flag pointing to where apache commons lang is on the server

I recall you disabled it automatically knowing where it is do to it causing a severe performance drop.

But Apache commons is an addition. The things that come with Java do work, yes?

Oh, yes

How can I test if those work with JDK 10? Any example?

(There are variables JAVA_HOME JDK_HOME JRE_HOME that you can set)

OK, I'll look into that.

I should get ready for work now.

@Dennis Any Java program at all needs to find the libraries that are the package java.lang.* so the normal Hello World should be fine

Ah, OK.

Hello World does seem to work. I guess the static build is smart enough.

@Pavel tio.run/#java-jdk (or tio.run/#java-openjdk9)

\o/ Thanks!

@Dennis Can you pull Add++ please?

@cairdcoinheringaahing Done.

@Dennis Thanks!

@Dennis Can you add dotnet script? (It's actually called dotnet script this time!)

It's the .NET Core variant of cs-csi and cs-mono-shell

That's third-party, yes?

@Dennis Yes

Well. It all seems to work! Great project, thank you.

@Marco I didn't use the install script btw. While I could have edited to match the desired location, I'd have to do so every time I update the language. I'm currently simply copying prelude.eg and the two .ego files in the sandbox's home directory, which also makes them discoverable.

@Dennis Mono on TIO is 5.8 but 5.10 has been out for nearly a month. Any idea why? 5.10 adds C# 7.2 support, which has some new features.

I probably need to dnf update.

You haven't done that all March?

Only for some essential stuff. dnf update has broken things in the past, so it's a manual process.

Oh, you manually specify which packages to update?

Huh

I should do that now.

Alright

Is it me, or TIO got way faster recently? An answer of mine that used to time out now runs in 8 seconds

Oh, that's fine. People won't notice. I find it awesome it works.

Well, I definitely noticed. It may be the difference between getting a time-out or the required output

I think that was a reply to chat.stackexchange.com/transcript/message/43549283#43549283.

@LuisMendo Not too many months ago TIO got some beefier multi-core CPUs

^^ Ah, that makes more sense

That was in January, yes.

It doesn't explain getting from >60 seconds to 8 though.

@LuisMendo Can you link the answer in question?

And it was >60 seconds on March 1 (this answer). Maybe I just got confused

^^ : ^

That's not 8 seconds, that's less than 2

I think case 3 used to time out. Perhaps it didn't

^^ It's test case 3

oic

A different version of Octave might make a huge difference. SymPy 1.0, e.g., is orders of magnitudes faster than SymPy 0.x.

So, did you update Octave this month?

Anyway, it's probably my mistake with the old timing

Since March 1? No, I don't think so. I'm updating everything right now.

The server could simply have been busy.

I think I found it. OP changed the test case. It used to be 8x8, now it's 6x6

Sorry for the noise!

That certainly explains it. ;)