Alright, first compiler flag. tio.run/nexus/…

So far only enabled for that language. More to come.

\o/

@Dennis What does -lm do?

Link the math library, which defines log.

@Dennis Also, possible bork: if you add a compiler flag, and it's empty, the wrapper fails.

That's expected. The same would happen if you executed it from the command line.

Ok.

Sweet, another thunderstorm, another blackout. Looks like it'll just be that language for now.

Ouch, must be annoying.

You get used to it. The part that irks me is that my phone has only 19% chargw right now.

Charge it from a laptop?

I tried to man bash and it froze my tab again ;-;

Feature request: get a better laptop. It doesn't even freeze on my phone...

The landing page should have a Minimum System Requirements :P

I wish that the debug info still had all that information like command being timed: and such.

I like understanding how TIO works inside, just a little bit.

The thing is that all the debugging info occupied a lot of space. 75% of the transferred data was debugging information... If you want to learn more about TIO, you can find the source code on GitHub.

Oh, I didn't consider the space it took up. You could add that as an option in settings: "Show extended debug information", which shows timing and all the rest. That way it doesn't fill the cache space and still allows anyone curious to look at it. If left unticked, Debug is just the program's StdErr

I agree btw, debug was too long previously but "command being timed" would be useful to check CLAs

@Dennis I go with "command being timed" too, especially if I want to use TIO for testing code that has specific time delays before submitting it to PPCG.

I'm not sure I understand your point. The time is still shown, just the command being timed isn't.

@Dennis Hi, Dennis! Any more news about this? Do you have to pass any further steps before you get the tenure?

Just a few months of agonizing wait.

:-)

Are there any other candidates that you compete with? I mean, how uncertain is it? And why such a long wait?

Also, for saving bandwidth, I don't think you need "CPU share".

@LuisMendo There's an opportunity every 4 years, so there's a bunch of candidates for each area (Analysis, Algebra, Geometry). I'm fairly certain I got it though. Looks like my wife did too. The long wait is because Paraguay.

@EriktheOutgolfer CPU share is actually quite useful for debugging, at least as far as maintaining TIO goes. If the CPU share is too low, something's wrong.

@Dennis That's great news! I won't say congrats yet then, but I'm glad the perspective is so good for you both

I think that it should depend on how many programs are being run at the same time, e.g. if 10 users use TIO simultaneously, CPU share should be equal, thus <10% for every user (because the system itself takes some CPU). Because queues are just irritating in overcrowding situations ;)

@LuisMendo And it's about time. Both of us would have won 4 years ago, but we both lost due to technicalities...

Oh. Then you will be even happier now! :-) Is your wife a Mathematician too?

Yep. We met at the master's course.

@EriktheOutgolfer It's a lot more complicated than that. Even a single request can wind up with a tiny CPU share if it has to wait for other resources.

@Dennis Like programs that run before the interpreter starts up and runs the user-specified code, such as timers?

Yes and no. Every running process competes for CPU time, but you may also have to wait for memory access, disk access, etc.

Well, you shouldn't fill the hard disk with useless stuff then, it only worsens the fact that access on a large disk is slow (e.g. zip files).

(usual tip, not to imply anything)

Compiler flags have been enabled for the following languages.

assembly-as
assembly-gcc
assembly-nasm
c-clang
c-gcc
cpp-gcc
c-tcc
intercal
java-openjdk
java-openjdk9
nim
objective-c-clang
rust

why is c-tcc lower than cpp-gcc on the list o_o

@Dennis Next feature: Add bytecount for compiler flags in case of compiled languages, arguments otherwise.

But it should not be universal: each compiler flag or argument must have an independent bytecount, so you can subtract uncounted bytecounts easily.

@betseg No clue. Ask sort.

@EriktheOutgolfer Waaay to many edge cases.

Next step are command line options for interpreted languages.

@Dennis Well, PPCG consensus is to count options that are not generally required to run a program, except for some specific exceptions. There can also be an option to count it or not. Then, there are two additional bytecounts, universal (the sum of all counted arguments) and general (code size + universal). Then, the PPCG submission header would say something along the lines of <l>, <b> bytes + <o>, where <l> is the language name, <b> is the bytecount and <o> is the counted argument count.

I understand it may be difficult to implement, and you have already set other priorities, and this is not really so major.

so i dont have to count -lm right?

That depends on PPCG consensus.

you have to include it to run the program, so...

Here I was talking about a feature not yet implemented.

As I said, waaay to many edge cases. I'm not saying it will never happen – and if someone wants to implement properly, I'll gladly accept the pull request – but there many things on my to-do list that are more urgent.

@betseg Assuming we're talking about the same answer, you have to count -lm (+4 bytes), but you don't need to include math.h.

oh

@Dennis Syms has been updated.

@CalculatorFeline Pulled.

@Dennis I'd like to add Crayon to TIO soon, but first I have a question about input: How does e.g. Jelly handle strings as command line arguments? If you give it a string as input, does it include the surrounding quotes when it's put in sys.argv, or do you have to put a backslash before them?

The reason I ask is because process.argv in Node doesn't seem to include the quotes.

@ETHproductions How are you calling it?

If I do crayon code.crayon "string", the second item in process.argv is just the string string.

Hmm, after playing around with Jelly a little, it seems to assume the argument is a string unless it looks like a number or array

That's because the shell processes the quotes. Arguments on TIO will appear verbatim in process.argv.

Oh, OK. So just to be clear, an argument of "string" on TIO gets passed to the command with the surrounding quotes?

Yes, exactly.

Jelly calls eval first and returns the bare string if it failed. tio.run/nexus/jelly#@//////oosS89FSNbE2FtPwihWyFzDwFiIihgWYsAA

OK, thanks. I'll double check my input process to see if it'll work as intended with TIO.

Jelly taking input as Python code is wierding me out a little.