Emscripten to run IOCCC entries on the website

[Buggem]

It'd probably be a welcome edition compared to having to have download the source and compile it yourself :)

And it would also allow easy access to running them. I know that Andy Sloane, donut.c creator has set up something simillar on his website (scroll to the Output section)

[xexyl]

Depends on how it works. (I don't know.)

As the security document says: run at your own risk.

I do know there is the thought about syntax highlighting though. Does it use JavaScript? That's one thing that the judges are trying to not use (or use as little as possible.). Also if it won't work with GitHub pages (not sure if that is what they call it) then that might be a problem too.

But of course right now the thing that is being worked on is the next contest. I know at least one judge is going on holiday after the contest too.

It's a curious idea though that perhaps you might want to make as an enhancement request after the contest? (Though starting as a discussion also is good - but if it is to be done it would be good to have it as an issue.)

Landon is focusing on one of the submission tools now but hopefully he will be able to react sometime soon!

I might add that some entries this probably won't work though as many are interactive and have a lot of different ways to invoke the program. Or is that something that can be done too?

Best wishes!

[Buggem]

Also if it won't work with GitHub pages (not sure if that is what they call it) then that might be a problem too.

It works with GH Pages, I've made a DOOM port in Emscripten hosted online here.

Does it use JavaScript?

The answer to that is somewhere between a yes and a no.
The program is compiled to WebAssembly, which is assembly but for the... well, web.

It's launched through the JavaScript WASM (Web Assembly) API, but the use of JS is minimal at the end.

I might add that some entries this probably won't work though as many are interactive and have a lot of different ways to invoke the program. Or is that something that can be done too?

Interactive entries will def be a challenge to implement, if possible at all without modifying the source code (removing the O from IOCCC). I could always hack up something, I'm alright at WASM-related stuff.

Thanks for getting back so fast 👍

[xexyl]

Also if it won't work with GitHub pages (not sure if that is what they call it) then that might be a problem too.

It works with GH Pages, I've made a DOOM port in Emscripten hosted online here.

Does it use JavaScript?

The answer to that is somewhere between a yes and a no.

The program is compiled to WebAssembly, which is assembly but for the... well, web.

It's launched through the JavaScript WASM (Web Assembly) API, but the use of JS is minimal at the end.

I might add that some entries this probably won't work though as many are interactive and have a lot of different ways to invoke the program. Or is that something that can be done too?

Interactive entries will def be a challenge to implement, if possible at all without modifying the source code (removing the O from IOCCC). I could always hack up something, I'm alright at WASM-related stuff.

Thanks for getting back so fast 👍

Sure. Well we'll see what happens after the next contest anyway.

But certainly you could open an issue later on and help out if it's wanted.

Have a great day and best wishes!

[xexyl]

Btw: some authors have videos of their entries on YouTube. It would be nice to have more demos if possible but not sure if it is. I guess we'll see.

[lcn2]

We are fairly sure that GitHub would want to host running IOCCC winners. They allow us to have an extensive web site because our repo is about source code, and we have workflow that helps manage the building of the web site, but to run winning code on their site would go way beyond their willingness to host our repo.

Some tools require extensive CPU and resources to fully compile. Some code deliberately crashes. Some code can kill other processes. Some code can consume substantial resources when running. Some code requests state between successive runs. Some code requires highly interactive infrastructure to use, etc.

This project looks like fun. And perhaps someone's servers can host an execution platform for some of the IOCCC winning entires that are amenable to running under such a platform.

If someone wants to try to provide an executable platform for people to try IOCCC code, they are welcome to try. And should such a project be implemented and is successful, we would be happy to provide pointers from this site (after IOCCC28 is completed and published).

[xexyl]

Very true. Many many entries are very interactive and some in ways that would be really hard if not impossible to have running on a website. Like 2019/endoh. Not only do you need to allow coredumps (which a lot of servers don't even allow, even worse with debug symbols in) but you have to then use gdb on it in fun ways. A brilliant masterpiece (no surprise from Yusuke).

I don't see how it could manage the brilliant 2018/mills either. It takes a lot of steps to set up. And same with other emulators.

Games would be hard to manage too and many other types of programs would be too.

It IS a fun idea but I would never personally run it on my server - it would be a security mare. I wouldn't speak for anyone else but I would be surprised nonetheless if there are (m)any who would be willing to do this on their server.

As for my entries I actually think it would be a shame to run on a website - no matter what you could not appreciate them in full (in fact very little of them). Videos of entries are nice (I guess) and I know at least two authors have videos of their entries but even some of those entries you can't get in a single video.

Yes I've given some thought to these things MULTIPLE times since my initial reply.

[lcn2]

Best wishes to somebody who wants to try and host execution of winning IOCCC entries. We don't have the resources to do it ourselves (we were spending about five euros a day on our infrastructure already).

Is somebody wanted to try and host execution of entries, we would suggest that they evaluate entries on a case by case basis, depending upon what the entry does and the capabilities of the hosting platform.

After IOCCC 28 is completed and published, and if there are regular links that we could use, we would be happy to add those links to our pages so that people could try such things out.

Best wishes

[xexyl]

Best wishes to somebody who wants to try and host execution of winning IOCCC entries. We don't have the resources to do it ourselves (we were spending about five euros a day on our infrastructure already).

I am very sorry for that and I wish I had the means to help out. Yeah I know I help out in other ways but I wish I could help in this way too.

Is somebody wanted to try and host execution of entries, we would suggest that they evaluate entries on a case by case basis, depending upon what the entry does and the capabilities of the hosting platform.

I've thought of that as well (as can be seen in my other reply).

After IOCCC 28 is completed and published, and if there are regular links that we could use, we would be happy to add those links to our pages so that people could try such things out.

And this should be a case-by-case basis too since some entries really should not be demoed on a website. But of course one could look at how it functions and then decide from there.

[mame]

Emscripten does not natively support CUI programs well enough. I am working on xterm-pty which connects Emscripten and xterm.js. I have confirmed that sloane/2006 works on xterm-pty. See the demo site.

However, it would be generally hard to run any IOCCC program with Emscripten for several reasons.

• Emscripten is clang-based. In fact, sloane/2006 requires a strange macro to force compilation through.
• Emscripten's runtime environment has so many incompatibilities with standard Unix/Linux+glibc that IOCCC's works are often not portable.
• Programs that animate, like sloane/2006, want to run on a Web Worker. However, for a technical reason, it is often challenging to run Wasm on GitHub Pages. This is because xterm-pty's demo site runs on netlify. To do it on GH Pages, you would need to use coi-serviceworker or something hacky.

Just FYI.

[xexyl]

Well okay: maybe I would be okay with it depending on what it looks like. But I am sceptical it would be good enough because there are so many features in my entries. In one it does something extremely different depending on how it's compiled (and there are many ways you can compile that one!). Actually another one is similar but the difference is not as significant even if it is different (well it is but mostly about colour schemes).

In fact there are THOUSANDS of ways that one can compile that first one. As for those two you mention I see you say it doesn't work but not sure why as they do work with clang. If they don't it's because of the system, not clang itself.

And looking at the link the problem is the code. That will NOT work with clang because clang mandates that the second arg is a char **. The first must be an int and any others must also be a char **. Furthermore some versions restrict the number of args to 0, 2 or 3.

The fix is in the code in the website.

THIS is how I fixed it:

--- sloane.orig.c	2024-12-30 15:02:34
+++ sloane.c	2024-12-30 15:02:34
@@ -13,7 +13,7 @@
        "~|_."[c(6,"b"              "cd\\a[g","^`"   "_e"
        "]fh")+8], p(   "#$%"        "&'()*+,-.1"[   c(11
       ,"_ac[]\\YZi"   "jkm",         "`bd^efghXWlV" )+13
-      ]-34,a);}main   (k,Z){         float z[1920]  ,A=0
+      ]-34,a);}main   (k,Z )char**Z;{float z[1920]  ,A=0
 ,B=0  ,i,j; puts(""         "\x1b"   "[2J");for(;;
 )  {  float e=sin(A         ),  n=   sin(B),g=cos(
 A),m   =cos(B);for(k        =0;k<   1840;k++)y=-k

UPDATE 0

Well if it works it's odd: because of what I said. That macro is not compatible with clang.

[xexyl]

I have to be quick (it's my birthday and I have family calling here in an hour or so and I have other things I want to do before then) but even though I am pretty sure my entries won't be good on a website I will provide you with some tips that might help you out.

0. Make sure you have the code from the repo as it has fixes (a fair amount by Yusuke Endoh actually but his fixes are gcc specific either because he didn't have clang, didn't realise it would be different or for some other reason[0])
1. For clang see the above comment. It is very strict on things. It also does not accept -traditional-cpp but I fixed entries that required that to work without it.
2. If you need to see what changed in a fix try make diff_orig_prog (there are other diff rules: see var.mk).
3. Sometimes a simple cast is all you need but other times you have to do more. Some of the early year entries I made show this.
4. Some will certainly not work. I mentioned one before - 2019/endoh is a backtrace quine for example. It deliberately crashes and needs a coredump with debug symbols in.
5. Sometimes disabling / enabling optimiser can fix issues. See what I had to do for one in 1986 here! That was the craziest one but one entry an author said it sometimes crash and it turned out to be due to the optimiser!

And if you have questions about fixing things I'd be happy to (within time constraints - I am both helping getting to the next contest and I might very well participate - plus there are other tools that need to be written for the contest) offer some ideas. You're welcome to mention me.

Finally if I sounded irritable I apologise. I probably was. Although it's been a great day I have been awake since stupid o'clock.

Best wishes! Hope these things help you out in your task.

[0] In macOS gcc is actually clang so this was easy to test (I could do it in linux too of course but then I have to do CC=clang). Like I said before some versions of clang are even stricter. That's why for some entries that only had one arg to main() I added a second unused one. Not ideal but it was important. Actually the version that complains about number of args seems to allow 1 arg only but I updated them just in case - for the C standard body as well as the compiler writers.

[Buggem]

Ok, you can take the project up... nobody likes a C newb like me...
It would be fine for an opt out website demo (eg some users can opt out)

[xexyl]

Ok, you can take the project up... nobody likes a C newb like me...

It would be fine for an opt out website demo (eg some users can opt out)

No that's not true at all! We all start somewhere. That's why I gave you some tips to work on this should you need or want. And I really do mean that I would be willing to answer questions.

As long as it's up to the author.

I don't want to discourage you. Like I said I have been awake for too many hours. So I am sorry you took it that way. Please do continue! It's a great way to learn actually and that is always a great thing!

I would say more but it's time for cake. Best wishes!

[xexyl]

As long as it's up to the author.

I don't want to discourage you. Like I said I have been awake for too many hours. So I am sorry you took it that way. Please do continue! It's a great way to learn actually and that is always a great thing!

Sorry. I was rushing. What I wanted to say was that I am sorry that I worded it the way I did. I was feeling a bit off at the time.

It is true that I think many entries won't work this way but that doesn't mean it's not worth looking into.

I DO see what you mean. If it can be worked out for some entries it might be nice. It could indeed make it even more presentable and that is a great thing.

I was just trying to finish something up before company got here for my birthday and so I worded it hastily and maybe a bit harshly.

I am very sorry. It was not intentional.

I believe that you should continue doing this. And as far as nobody liking a newbie that's not true.

I will impart some advice that a friend of mine who gave me the opportunity to get really good at C imparted on me.

He said: there's no such thing as a stupid question; there are only stupid answers. Your idea is not a question but the idea is similar.

Also when I was really tired and made a bad mistake (that caused memory corruption that took some time to locate due to the part of the code being long past due to it not going way out of bounds) he told me a saying in the Netherlands (he's Dutch) that loosely translates to: where there's woodwork there's wood chips: which basically means that when we do things we're bound to make a mess of things. Which means that even of this doesn't work out (but maybe it will) it still is worth trying. You will learn a lot and that's a great thing.

So again I am extremely sorry. I actually encourage you to continue this. And who knows - maybe you can participate and win the IOCCC one day!

Anyway I need some sleep. I just felt the need to properly apologise and to express that you should not be discouraged by someone. Just because I am not sure I want my entries like this doesn't mean it's not a worthy endeavour and depending on how it's done it might even be really great.

Best wishes!

[xexyl]

Thinking about this I am reconsidering it a bit. I also have an idea that (depending on your original idea) might be simpler (and another that might be helpful in another way).

I think the real trouble is the interactive ones and the ones that are meant to crash or play with the system[0] (or that have to be run in a terminal or terminal emulator etc.). As for mine I think one might work but the others are harder and one extremely hard to do (at least to show all they have).

As for the idea. Would asciinema work? It records so it's like the demo you shared in the OP. Also many entries have one or more scripts that can be useful (many compile the entry in fact).

Finally (and this is all I'll say on the matter) I am sorry again. Sleep has been rubbish (worse than usual), I've been far more tired than usual and I've been super busy. Not an excuse for being harsh but it is what happened.

If you can get this working it might very well improve the presentation of some entries and although I might not be able to answer everything and I might be slow at times I am happy to help where I can.

Whatever the case don't ever think you're not worthy because of [anything]. Everyone has something to offer and sometimes the people who are the least experienced (of all) can come up with the most brilliant solutions to problems! There is science behind this but that's not really relevant. Keep it up whatever the case.

Best wishes.

[0] One in fact messes with the screen and interferes with the cursor in an obnoxious way. Another can if you're not careful do what amounts to a fork bomb! That would be problematic.

[Buggem]

Emscripten crashes are close to crashing the whole browser. So forkbombs... must be handled... carefully.

[xexyl]

Emscripten crashes are close to crashing the whole browser. So forkbombs... must be handled... carefully.

That's unfortunate indeed. As for the fork bomb one it might not in current systems. I haven't actually checked it and system limits (ulimit etc.) possibly protect it more now. The entry I refer to is old. But 2019/endoh is supposed to crash and the user must (to use it and enjoy it) allow coredumps. Not only that but it has to have debug symbols in it (so no optimisation either). As you might know some admins do not allow coredumps and some especially do not like debug symbols included (though I would hope that anyone who is running an IOCCC entry actually is not running such an entry on a critical system :-) ). There is of course good reason for this but that's another matter entirely.

There are a lot of different kinds of interactive programs. Besides the obvious (games - including one of mine - Snake, 2020/ferguson1) there are shells, system emulators, one that uses the shell to answer questions (in a fun way) and various others.

Some are pretty simple and would easily run if you can get it working. If you do go for this for 1984/mullender you'll have to use the alt code (that I added a while back) - the original will only work on VAX-11/PDP-11 systems (it is Landon's all time favourite entry even after all these years). But there is something subtle about that one: you might not realise it but if you press enter when it reaches EOL it will continue again.

And that's the thing. A lot of entries (all of mine but many others too) have a lot of subtle things that are hard to determine. Some of my entries have some extremely subtle things in them that are very easy to miss.

The one that has thousands of ways to compile is actually an understatement. Yet that very claim depends on how you look at it. The thing is it allows you to change variables at compile time and those variables can be a huge range. But different values will cause different things to happen and the program has a couple Easter eggs too. I won't reveal how that works but as you can see there are a lot of things to consider.

But that doesn't mean your idea doesn't have merit. It just has to be done in a well enough way and give the author the option to opt-out. If you can get it working then great. Some certainly won't work but if it gets more people interested in it that's GREAT!

I am kind of a purist at times and I am a pedant and a stickler. Those are known faults of mine (sorry!). There have been things released out there that have got people interested in something but the released thing did it poorly and so there is a question of: is it a good thing or a bad thing? I'm usually inclined to say it's a bad thing but there is another argument and in some ways it might be good (it might actually get some people to get into it enough where they can appreciate the view that it's bad!).

Anyway I have my cousin calling here (as in visiting) and she'll be here soon so I have to go.

BTW: I don't know anything about WASM and I generally (though you might not think it from my work on this website) don't like web design. Right now I'm working hard on getting it so IOCCC28 can open and I have other things (other tools) I have to do after it.

Whatever the case: keep up the GREAT WORK!

Cheers .. BCNU.