talk.tryitonline.net

OK, but my point is that things change over time.

Python 2.7 started to segfault because a newer gcc aligned a struct differently.

@Pavel Do you have 32-bit development files?

Pavel

Good point

Oi, why does install 32 bit libc downgrade 64bit gcc

dnf wat

Yeah, no. the configure script just doesn't pick it up

None of these seems to contain your C library.  What is the full name
of your C library? /usr/lib/libc.so

Extracting names from /usr/lib/libc.so for later perusal...ar: /usr/lib/libc.so: file format not recognized

The archiver doesn't think /usr/lib/libc.so is a reasonable library.
Trying nm instead...
nm: /usr/lib/libc.so: file format not recognized
That didn't work either.  Giving up.

Ah hah, I ffigured it out

It got confused by the link from libc.so to libc.so.6

> make: *** No rule to make target '<built-in>', needed by 'arg.o'. Stop.

Well that's familiar

But no type errors

2:50 AM

@Pavel tio.run/#calc2

Speaking of calcs. github.com/lcn2/calc

I'll take a look at that one tomorrow.

3:10 AM

@Dennis whats the path for the crt on TIO?

@Downgoat crt?

3:30 AM

what is weird is that TIO Clang doesn't seem to support the standard -arch flag tio.run/##S0oszvj/…

and I can't find a way to detect which Clang installatoins support it and which dont

aha found it!

-target?

Does LD have a cross-platform way to specify arch?

4:07 AM

@Downgoat Not architecture per se, but ld -m elf_x86_64 implies one.

4:24 AM

> ld: warning: option -m is obsolete and being ignored

4:36 AM

@Dennis I've redone the toolchains system so should work now. Could you pull VSL?

4:51 AM

uhh i take that back still might be buggy unless clang is installed

1 hour later…

Pavel

6:18 AM

@Dennis Thanks! You didn't push tryitonline.

7:09 AM

@Dennis could you update ngn/k please

1 hour later…

8:16 AM

ok I think I've squashed the last few bugs VSL should be good to go tested on macOS ubuntu and ubuntu

5 hours later…

12:57 PM

@Downgoat What do you mean? TIO does have clang istalled.

@Pavel Fixed.

@Downgoat It's working now. Updated.

@ngn Doesn't seem to pass it's self test.

testing
-1 23 -45 -0 0N 0W -0W
 expected: -1 23 -45 0 0N 0W -0W
 actual: -1 23 -45 0 0N 9223372036854775807 -0W
-1 0 1%0
 expected: -0W 0N 0W
 actual: -0W 0N 9223372036854775807
(+/!0;*/!0;&/!0;|/!0)
 expected: 0 1 0W -0W
 actual: 0 1 9223372036854775807 -0W
&/0#,(1 2;"34")
 expected: 0W
 actual: 9223372036854775807
$(0W -0W;0N -12345678;9)
 expected: (("0W";"-0W");("0N";"-12345678");,"9")
 actual: (("9223372036854775807";"-0W");("0N";"-12345678");,"9")
5/345 failed
'error
$[n;error+1($n),"/",($#t)," failed\n";];

@Dennis huh... could you give me the first line clang -v prints?

clang version 6.0.1 (tags/RELEASE_601/final)

@Dennis mine is way older (3.8.1). i'll try to figure out what's going on and i'll ping you later today

2:08 PM

@Dennis it should work now. the problem was this

2:45 PM

OK, I'll pull as soon as I get home.

3 hours later…

6:14 PM

@Dennis could we make the TIO debug/output textfields parse ansi color codes :3

7:06 PM

That shouldn't be too hard, I think.

@ngn Updated.

@Dennis thanks!

@Dennis uh... seriously? o_o (although it can be harder for other escape codes)

Emulating all escape codes would require some work, but \e[31m just has to be mapped to  or something like that.

more like \x1b\[31m, tbf

(regex)

What is all that stuff at the front?

7:19 PM

sorry

I had a moment where I thought \e was a literal \e lol

(did some serious math today)

so, yeah, I guess you can do that

although it might lead to unclosed tags

If you map it to , that won't be a problem.

ah, so you prepend  and append  and then do that?

Basically.

Of course, the easiest solution would be to render the output in a virtual terminal and take a screenshot.

that would make the output uncopiable

It's also neither easy nor possible inside the sandbox.

2 hours later…