« first day (868 days earlier)      last day (1626 days later) » 

12:29 AM
Installed it from Forge. The system package seems to be broken.
 
12:47 AM
Why is there 5.1 GiB of stuff in /root?
 
Interesting question
 
3.4GiB are being occupied by stack.
Why do I need that again?
 
What stack?
 
It's a build tool.
git blame blames PureScript.
 
If it's a build tool you could probably remove it after build
 
12:50 AM
2 days ago, by Dennis
PureScript now requires a new build tool. sighs
Hey, I had fun with PureScript too when I upgraded to Fedora 28.
May 17 '18 at 15:08, by Dennis
So you cannot install Purescript as root because the home directory of some other user isn't writable...
@Pavel If I want to update it, that's probably not the best idea. PureScript takes half an hour to install, and I suspect creating that giant directory is to blame at least partially.
Speaking of things that take time: how long should the cpanm command take?
 
@Dennis Normally maybe 10 seconds or so. For how many modules Task::Kensho installs, much much longer.
But not necessarily longer than your regular cpan command
 
1:06 AM
It's been running for over 30 minutes...
 
I've had cpan commands take that long
Well as long as it's still producing output it's probably doing something
 
I suppose this would haven taken a year or so with cpan.
It's doing something, but I'm not excited about adding another hour to the build time. It already takes more than four...
 
Task:Kensho is broken into submodules: metacpan.org/pod/Task::Kensho
Task::Kensho::Async is probably useful but Task::Kensho::Email wouldn't be
cpanm Task::Kensho::{Async,CLI,Config,Dates,Exceptions,Hackery,Logging,ModuleDev,OOP,X‌​ML}
Of those, CLI, Logging, XML would work on TIO but not be very useful. ModuleDev has a few useful packages but it might make more sense to install them seperately, ModuleDev contains a lot of stuff for uploading to CPAN
Data::Printer, Devel::Confess, Perl::Tidy
 
1:30 AM
So, which ones should I install?
cpanm Devel::Peek Moose Task::Kensho has been going for an hour now, so I'll interrupt it.
 
1:41 AM
Dangit, I forgot to sync after fixing Octave.
And I was going to restore /usr/local to undo the cpanm command...
 
1:56 AM
Shit.
New server should be back up.
@Pavel I have to stop for today. I already rebooted the main server by accident, and I don't want to do something worse.
@Downgoat I've updated the OS. TIO now has LLVM 7.
 
Awesome woohoo!
 
@Dennis I'd say OOP, Hackery, and Exceptions are the ones most useful for TIO. OOP also includes Moose.
 
2:13 AM
@LuisMendo It seems like symbolic no longer produces stray output with sympref quiet on, so I'm no longer executing catalan before matl, reducing startup time by roughly a second. Please let me know if this causes any issues.
 
Generally speaking, it's Perl culture to consider CPAN not to be separate from core modules and that dependencies can be added for anything.
 
@Pavel So, cpanm Devel::Peek Task::Kensho::{Exceptions,Hackery,OOP}?
 
Yeah
 
OK, I'll try that tomorrow.
I'm spent.
 
 
11 hours later…
1:01 PM
@Dennis Hey, thanks for noticing that! Much faster now :-)
 
It kinda forced me to notice. ;) My previous workaround for symbolic's stray output to STDOUT was to delete the first lines of output. That output was gone, so the Hello World printed nothing...
2 days ago, by Dennis
MATL had the easiest fix yet, but it took me forever to figure it out. ._.
 
@Dennis Oh. Sorry it gave you trouble :-)
How did it occur to you to try Hello World again?
Strangely, Octave 4.2.2 on Windows 10 still produces a line of stray output with sympref quiet on. It must be an OS dependent thing
 
I just upgraded to Fedora 29. Started Wednesday, finished yesterday.
 
Ah, I see
Upgrading the OS seems to have cost a lot of time then!
 
Well, 39 langauges were no longer working...
Most issues weren't actually related to the update. Outdated build scripts and whatnot.
Symbolic has been updated to 2.7.1. That's probably why the output changed.
 
1:10 PM
Ah. My Symbolic is 2.6.0
 
@Pavel That's not working at all. cpanm Task::Kensho::Hackery fails because Term::ProgressBar::Simple is not installed, cpanm Term::ProgressBar::Simple fails because Devel::Peek is not installed. But I already did cpanm Devel::Peek and that just printed skipping S/SH/SHAY/perl-5.28.1.tar.gz...
Any reason I can't just dnf install perl-Task-Kensho-*?
 
1:55 PM
@Οurous I think I need a lawyer before adding Miranda.
> Please note that rehosting or redistribution of any of the Miranda executables requires specific authorization and is not permitted otherwise.
> LICENSOR grants customer a non-exclusive non-transferable right to use the licensed software [...] on a single computer belonging to or for the time being under the exclusive control of customer
> Customer agrees to use the licensed software solely on the licensed computer and not to release or transfer it to any other person.
There's an email address for queries about licensing. That's probably cheaper than a lawyer...
 
2:14 PM
Can't get it to print Hello World anyway. I thought [Stdout "Hello, World!"] would work, but it says syntax error - unexpected end of file.
 
2:36 PM
@MilkyWay90 I don't understand how to use your interpreter. Please provide a Hello World and step-by-step instructions to run it.
Preferably in the repo. You should really provide at least minimal docs for your languages.
 
2:50 PM
@Dennis You probably can
 
3:21 PM
@ConorO'Brien tio.run/#rockstar
2
@Skidsdev ^
@Bubbler ^
@AncientSwordRage ^
@Nick ^
 
:D
 
wow that's a lot of people who wanted Rockstar there :-)
 
3:34 PM
So one of the things I was gonna do this week was get in touch with tio.net and see what you needed from us to get Rockstar supported... and you beat me to it. AWESOME. :)
(and thank you! and tio.run is really, really cool - nice work.)
 
honestly amazing it hasn't fallen apart yet no matter how much people designing build systems try to keep it from functioning
 
@DylanBeattie My pleasure. Please let me know if there are any issues. There were a lot of implementations to choose from; I went with rockstar-py.
 
3:55 PM
Yeah, that works pretty well for now. Work is in progress on an 'official' reference implementation, currently built in Python, so we'll let you know when that's ready for use. But rockstar-py should be fine in the meantime.
 
Alright. :)
New language: tio.run/#cpy
 
4:15 PM
Man I kinda wish I could see the intermediate C
I'm intrested how the example return sum, subtract, multiply, divide functions
 
:+1:
Ah, right, C++. Has tuples.
 
4:45 PM
@Dennis thanks!
 
4:57 PM
"New" language: tio.run/#rebol
 
@Dennis as with all of my languages, Sum-It is on hold for now. But i might have a new language for you by the end of the day :P
 
With any luck, I won't have backlog by then.
 
it'll be a simple language, as with all of my Node-based langs it'll be node script.js -f filename
 
That's reassuring. :P
 
Also fun fact: Every language I've made since switching from Python to Node has been made in TIO
 
5:11 PM
Why would you switch from Python to Node
3
 
I prefer JS over Python personally
Coming from a C#/PHP background, JS is more intuitive to me. Also coming from a web dev career I actually use JS day-to-day
 
Why not stick with C#?
C# is good
 
^
 
it is, I love it
 
I'm not sure if Node or C# has caused me more headaches. :P
 
5:12 PM
it's also very verbose and relative to JS takes a lot of boilerplate for quickly prototyping things
Also the way I tend to make languages doesn't work well in C# due to its strong static types
 
C# has dynamic
 
As long as you don't have any dependencies, Node should work as well as anything else. But if you do, you need npm...
 
I very rarely use any dependencies when making languages
usually the only require I have is require(fs)
 
And for that reason, adding recreational languages is usually a lot easier.
 
F# is good too
Imagine a less verbose, functional C#
 
5:22 PM
@Dennis You're gunna hate me when I eventually finish NuStack
it's a C# Lexer/Parser that outputs NASM code
originally, for some unholy reason, it was a C# lexer that output tokens, then passed to a JS parser that output NASM
 
TIO has nasm
 
None of that sounds too scary.
 
I can just imagine it being a pain in the ass having to pipe the outputs around like that
 
Unless you wrote it for .NET Core. Then it's a bit scary.
 
It is
 
5:25 PM
You could always include a wrapper in your repo. ;)
But writing a wrapper isn't hard. Getting npm to function as intended and not brick your VPS... that's another story.
 
I suppose it'd be fairly trivial, even back when it had the JS component, with something like dotnet Lexer.exe myFile.ns | node Parser.js | nasm
@Dennis what about NuGet?
 
Not sure how you got netcore to output an exe
We've been trying for a long time
 
you know what I mean
 
We can't dotnet run in a TIO wrapper
 
oh?
 
5:27 PM
Yep
 
it's fairly trivial C# code tbh, I'm sure with few if any modifications I could make it work with mono as well
 
We can't?
 
@Dennis We tried with Broccoli, didn't we?
 
We probably did.
 
I have a feeling it tried to restore nuget packages and fell over but I don't remember
 
5:28 PM
@Pavel That was so much faster. Installed.
 
\o/
 
5:50 PM
So I'm working on a simple language that basically uses single-char tokens as substitutes for JS snippets, and uh.. this is hello world =a*'ua++a*", "+a*ua,+a*"!"pa*
wait no, I'm just bad at golfing my own language :P p=a*+++'ua+", "ua,"!"
oh wait =a* isn't needed either :P
 
6:06 PM
Failed to process most of the requests from the backlog, so I guess I'm done for now.
 
 
1 hour later…
7:14 PM
@Dennis I always forget to read the licence. And also that licences are important to TIO. Thanks for checking /looking at it.
 
7:26 PM
@Dennis - I was looking at COMAL again; as near as I can see, the implementation is an interactive terp; I'm not sure if OpenCOMAL's "runtime" can take text-source as an input rather than a tokenized SAVE from the terp. Does this cause problems with respect to getting it up on TIO?
 
Hard to say without knowing more. We have to separate code and input somehow; we can't provide both on STDIN.
Somehow overlooked COMAL when I was going through the backlog. :/
 
7:47 PM
OK, just tried the MSDOS build of the josvisser.nl package; the comal runtime (OCOMRUN.EXE) definitely doesn't take a textfile as the "run source"; it has to be a tokenized file as generated by the terp SAVE command. And I haven't seen any indication that the terp can be operated in "batch" mode, where one might be able to build a wrapper that would load the terp to read the user's source, write a SAVE file, and exit, then invoke the runtime to read the SAVE file and execute the user code.
A bit sad in a way; COMAL could have been an alternative to QBASIC; it addressed pretty much all of the criticisms of GWBASIC, in ways amazingly parallel to what eventually became QBASIC.
 
 
1 hour later…
8:58 PM
@Dennis Told you I'd have a new language by the end of the day!
 
@Skidsdev tio.run/#clam
 
Aww you were faster than I expected, I'm just about to finish up and push while loops :P
@Dennis ok there we go, can you pull it again please? :)
 
9:13 PM
Done.
 
(Don't worry, after this pull I won't be working on it more until tomorrow)
Thanks
 
 
1 hour later…
10:37 PM
New language: tio.run/#bitch
 
10:50 PM
@Dennis Ok I know I said I wouldn't work on it until tomorrow, but there were a couple lil bugs I needed to tidy up, could you pull Clam again please?
 
@Skidsdev Done.
 
Thanks
 

« first day (868 days earlier)      last day (1626 days later) »