« first day (519 days earlier)      last day (1975 days later) » 

12:36 AM
@cairdcoinheringaahing v
File "/opt/addpp/add++.py", line 8
    import errorimport functools
                               ^
SyntaxError: invalid syntax
 
lol. I think you forgot a newline there, @cairdcoinheringaahing
 
12:53 AM
@Οurous The easiest solution seems to be to replace the new versions of ArgEnv, Generics, and StdLib with the deprecated ones. There's probably a good reason to avoid this, but after rm -rf ArgEnv/ Generics/ MersenneTwister/ StdLib/; mv Platform/Deprecated/* ., there are only the following errors:
 1  Error [ESMVizTool.dcl,4]: iTasks.dcl could not be imported
 2  Error [ESMVizTool.GraphvizVisualization.dcl,4]: iTasks.dcl could not be imported
 3  Error [Text.Parsers.ZParsers.MetarDemo.MetarParser.dcl,1]: module name "Text.Parsers.MetarDemo.MetarParser" does not match file name: "Text.Parsers.ZParsers.MetarDemo.MetarParser.dcl"
 4  Error [_library.dcl,2]: incorrect module header
 5  Error [_startup.dcl,2]: incorrect module header
 6  Error [_startupProfile.dcl,2]: incorrect module header
 7  Error [_startupTrace.dcl,2]: incorrect module header
ESMVizTool requires iTasks, Text.Parsers.ZParsers.MetarDemo.MetarParser is a regression, and the files with _ probably aren't meant to be imported in the first place.
 
1:04 AM
@Dennis There's no problem with using the deprecated ArgEnv, and the deprecated Generics just references Data.Generics so there's no problem there either (until the new Generics is finally finished off) - the deprecated StdLib however, is missing a bunch of fusion macros and list/array extensions from the new one.
From what I've seen though, if you swap ArgEnv and Generics, compile everything with Platform/Deprecated/StdLib, then compile it again with StdLib, the *.abc files can work with references to the new StdLib even if they needed the old one to be generated properly.
And yeah, underscore files with lowercase names aren't for importing
 
1:18 AM
@Dennis Can you pull Attache please ?
 
1:39 AM
@ConorO'Brien Done.
 
thank you!
 
@Οurous Another option would be to not pre-compile anything, copy /opt/clean/lib to a user-modifiable location, and let everything compile naturally from within the sandbox. I'm not sure how much slower Clean would become, but that should solve this issue once and for all, no?
 
Why would moving /opt/clean/lib be better than, say, chown -R runner /opt/clean/lib?
 
Copy, not move. 1. runner can't modify files from within the sandbox, even if they're owned by runner. 2. Allowing sandboxed programs to permanently modify files in /opt/clean/lib is a major security risk.
 
2:02 AM
@Dennis It would solve this issue (with compilation errors) in a general sense, but it would introduce others as well as rendering the current StdLib and Data.Generics unusable together, which (given that being the reason for the consideration of not precompiling) is undesireable.
 
@Dennis can you add MoonScript?
And Yabasic if possible, please?
 
I can formulate an exact, step-by-step pre-compilation process based on my linux install if that would save time?
 
luarocks install moonscript and rpm at yabasic.de/download.html, respectively.
Moonscript looks cool
 
@Οurous That would be great. When testing different compilation methods, all I can tell is whether I see compiler errors or not. My limited Clean knowledge doesn't really allow me to test the compilation results...
@MDXF Added to the list.
.oO(That's a long list.)
 
Thanks
Do you have an online copy of the list we can all see?
 
2:07 AM
I do not.
 
Hmm.... Pastebin it?
 
If I wanted to make TIO's to do list public, I would. People tend to confound plans with promises...
 
Ah I see
 
Hold on, I think I've seen the moonscript logo somewhere before... nah.
2
 
lol
 
2:16 AM
Also @Dennis do you mind pulling Cubically please?
Oh wow I've optimized the interpreter so much since TIO's Cubically has been updated, ~3 million more turns per second best-case since commit d406ed
 
2:30 AM
@MDXF Done.
 
Thanks!
 
3:06 AM
@Dennis Simplest set of instructions I could come up with (paths are in reference to the root of the library directory):
rm -rf Generics
rm -rf Sapl
rm -rf iTasks
rm -rf GraphCopy
rm -rf Gast
rm -rf TCPIP
cp -r Platform/Deprecated/Generics Generics
rm -rf Platform/Deprecated/Generics
rm -rf Platform/Deprecated/ArgEnv
and then clm -dynamics -IL Dynamics -IL Generics -IL Platform -IL ArgEnv -IL Platform/Deprecated/StdLib -IL StdLib -IL Directory main, using the repeat-for-each-file-until-it-stops-changing method you currently use
 
Why remove Sapl, GraphCopy, Gast, and TCPIP?
 
TCPIP: Does TIO allow outbound connections?
 
It does not. So it's simply because we don't need it?
 
Yep. Gast = Generic Automatic Software Tester, for generating edge-case unit tests and powering the stagnant visual debugger
Although I guess we don't need to remove Gast / Sapl / GraphCopy, because they will actually work on TIO unlike TCPIP. And if someone writes another language in Clean, they'll be useful for the runtime.
 
3:35 AM
@Οurous That lets me compile everything except some deprecated stuff, ESMVizTool, and Text.Parsers.ZParsers.MetarDemo.MetarParser. However, nothing in lib/StdLib is getting compiled, because the stuff in Platform/Deprecated/StdLib has priority.
 
Ah. I'll put together a source file to force it to compile StdLib.
 
That might not be needed. I can modify the switches depending on the directory we're compiling.
 
@Οurous is Clean your lang?
 
@Dennis from StdLib import group
the symbol is only present in the new StdLib
@MDXF In the sense that I use it frequently, and enjoy doing so, yes. But no, I didn't create it (it's older than I am), and I don't contribute to the development (the indev version is closed-source, because of the theoretical/academic nature) even though I'd like to. (even the nightly is up to 6 months behind the actual master source)
 
@Οurous Will that compile all of StdLib?
Also, do we need the deprecated ArgEnv? The normal one seems to work just fine.
 
3:51 AM
I did rm -rf Platform/Deprecated/ArgEnv
 
Ah, sorry. Misread that.
 
And yes, it does compile all of StdLib, but for some reason it puts it in the folder with the Deprecated one
In that case, additional step: Compile StdLib with and without the -IL Platform/Deprecated/StdLib
Which makes the compilation steps "compile everything with clm -dynamics -IL Dynamics -IL Generics -IL Platform -IL ArgEnv -IL Platform/Deprecated/StdLib -IL StdLib -IL Directory main", and then "compile from StdLib import group with the same command", and then "compile import StdLib with clm -dynamics -IL Dynamics -IL StdLib main
Together, that should compile everything important, and make references to the deprecated StdLib compatible with the new one
 
4:18 AM
@Οurous Didn't pay attention to chat and tried the switch modification.
            if [[ $f =~ ^Deprecated|^ESMVizTool ]]; then
                continue
            elif [[ $f =~ ^Data/Generics|^Internet/ ]]; then
                stdlib=Platform/Deprecated/StdLib
            else
                stdlib=StdLib
            fi
That seems to work. Only MetarParser errors and StdLIb has its object files.
 
Yeah that looks like it'll work too.
 
4:35 AM
@Οurous I've synced the arena servers. Please tell me if there are any issues.
 
@Dennis Will do. Looks good so far, thanks again!
 
Thanks for your help.
 
 
11 hours later…
3:40 PM
@Dennis Fixed, could you pull it please?
 
@cairdcoinheringaahing My Hello World test prints Hello, World!\nHello, World! now.
 
Huh, it worked for me the last time I checked. I'll take another look
 
3:56 PM
@Dennis Should be fixed now, I had pasted in the new code without removing the old code on GitHub :/
 
Pulled.
 
4:35 PM
@Dennis Can you look at adding es some time?
 
Holy backlog, Batman!
4
 
 
1 hour later…
5:57 PM
Tfw you have to compile GMP from source because the system version breaks things...
 
6:09 PM
What uses GMP?
 
GAP
Nevermind, building GMP does nothing to fix this.
@MDXF Which GAP packages do you need? Some of them fail to compile.
 
6:24 PM
Clear notifications of previous executions once the run button is hit again.
 
@Laikoni You mean like if there was a bork and it diplayed the "Results could not be retrieved" notification?
 
@Laikoni That sounds like a good idea.
 
E.g. running Try it online! produces the warning that the output is truncated. Changing the program to main=print[' '..'~'] and pressing run again produces a non-truncated output, but the warning is still displayed.
 
Yeah, I've been there before.
 
6:39 PM
Compiling with --with-gmp=no still shows the same errors. Wtaf?
 
 
2 hours later…
8:13 PM
@MDXF tio.run/#gap Couldn't build all packages though.
 
8:34 PM
@Laikoni Done.
 
@Dennis Thanks!
Though you are spoiling us with those insanely fast updates.
 

« first day (519 days earlier)      last day (1975 days later) »