|
cum pilers
|
#
?
Jul 8, 2017 11:45
|
|
|
|
| # ? Apr 25, 2024 16:48 |
|
|
sick but fair
|
|
#
?
Jul 8, 2017 11:49
|
|
|
pram posted:cum pilers -Oface
|
|
#
?
Jul 10, 2017 09:25
|
|
|
pyf compiler bugs https://reviews.llvm.org/rL307092 quote:If the output file is a character file (e.g. /dev/null) unlinking it could thansk
|
|
#
?
Jul 10, 2017 18:39
|
|
|
that sounds like an os bug to me
|
|
#
?
Jul 10, 2017 18:48
|
|
|
yeah it seems like you shouldn't be able to unlink /dev/null without root access.
|
|
#
?
Jul 10, 2017 18:50
|
|
|
MALE SHOEGAZE posted:yeah it seems like you shouldn't be able to unlink /dev/null without root access. you can't. that purpose of that change seems to be about making it possible to run the build script as root. not sure why that's a desirable thing to support.
|
|
#
?
Jul 10, 2017 19:06
|
|
|
it looks like you can't and the bug is specifically about making it safer to run the linker as root? what the gently caress
|
|
#
?
Jul 10, 2017 19:06
|
|
|
TOPS-420 posted:pyf compiler bugs https://mobile.twitter.com/bos31337/status/116372971509121025
|
|
#
?
Jul 10, 2017 19:18
|
|
|
working as intended
|
|
#
?
Jul 10, 2017 19:19
|
|
|
https://twitter.com/johnregehr/status/695271704982069248
|
|
#
?
Jul 10, 2017 19:24
|
|
|
do I know anyone stupid enough to run the linker as root 🤔
|
|
#
?
Jul 10, 2017 20:20
|
|
|
The Management posted:do I know anyone stupid enough to run the linker as root 🤔 uh literally anyone running it in a docker
|
|
#
?
Jul 10, 2017 20:22
|
|
|
i'm pretty sure anyone that run 'sudo gem install blah' is probably running compilers/linkers as root
|
|
#
?
Jul 10, 2017 20:26
|
|
|
I'm happy I don't program in C or C++
|
|
#
?
Jul 10, 2017 20:39
|
|
|
I love the (unsigned)-1.0F is a random value, that must have been incredibly frustrating to track down
|
|
#
?
Jul 10, 2017 20:43
|
|
|
let me rephrase that. I know a lot of idiots running compilers and linkers as root
|
|
#
?
Jul 10, 2017 20:44
|
|
|
if you've ever typed sudo make install without running make first you've probably run a linker as root. it's obviously not something you should be doing, but it's not at all hard to do accidentally
|
|
#
?
Jul 10, 2017 22:13
|
|
|
Sweeper posted:I love the (unsigned)-1.0F is a random value, that must have been incredibly frustrating to track down Honestly that is a lot of screams terrible developer.
|
|
#
?
Jul 10, 2017 23:23
|
|
|
DuckConference posted:the volatile write was to a global variable, right, and not to something local like a pointer you declared locally within the function? JawnV6 fucked around with this message at 08:42 on Jul 11, 2017 |
|
#
?
Jul 11, 2017 03:49
|
|
|
why is it bad to run the linker as root?
|
|
#
?
Jul 11, 2017 07:37
|
|
|
programs really should not be run as root unless they've been designed to be extremely cautious about how they interact with the system. the average toolchain is no more designed to work that way than a word processor is i mean, it's not really a security hole on its own, not unless you're doing something that should set off warning bells in anyone not completely oblivious, like writing a compilation server or making the compiler setuid. but it does create a lot of unnecessary potential for bugs to completely gently caress over your system
|
|
#
?
Jul 11, 2017 08:48
|
|
|
spent two weeks on my own hobby compiler trying to figure out why it's stack implementation returned 10006 for (5000 + 3 + 2) turns out the assembler had a destructive check in an inner loop instead of an outer loop when evaluating index-plus-offset (like stack[ebp - 4]) variables it would get emitted as stack[ebp - 4] once, then as stack[ebp] for the rest of the uses of that variable holy poo poo i was so mad at myself when i finally figured it out
|
|
#
?
Jul 11, 2017 09:13
|
|
|
compilers are fun. i wrote at one at work a couple of years ago for an embedded scripting language and learned a lot doing it. the feature i'm the most proud of is the nice error reporting system which underlines errors with token-level granularity and gives sensible human-friendly messages. at the time, this was better than any real compiler we used. i found several fun bugs during development. my favourite was a performance issue where the compiler seemingly got exponentially slower the longer the input text was. after a few hours of profiling which didn't really help, i finally noticed that i had written: C++ code:
|
|
#
?
Jul 11, 2017 13:26
|
|
|
Spatial posted:one of those classic coding moments where my entire day's work was changing one character for a thousand times performance improvement. more like one of those moments when you realize why C++ is a terrible language
|
|
#
?
Jul 11, 2017 16:06
|
|
|
what's the over/under for how long before BDCE gets too clever and exposes a timing attack
|
|
#
?
Jul 11, 2017 18:49
|
|
|
p sure that's already happened. timing sensitive code these days more or less has to be hand-rolled assembly language
|
|
#
?
Jul 11, 2017 18:51
|
|
|
The Management posted:more like one of those moments when you realize why C++ is a terrible language still copy and pasting code with #include in 2017! 
|
|
#
?
Jul 11, 2017 20:01
|
|
|
TOPS-420 posted:p sure that's already happened. timing sensitive code these days more or less has to be hand-rolled assembly language I don't know who u are but sick username
|
|
#
?
Jul 16, 2017 18:10
|
|
|
as a millennial with limited attention span I can only use interpreted language s like Python
|
|
#
?
Aug 4, 2017 06:14
|
|
|
you should try java, it will teach you how to actually program and get a real legitimate job
|
|
#
?
Aug 4, 2017 06:16
|
|
|
JewKiller 3000 posted:you should try java, it will teach you how to actually program and get a real legitimate job Or write lines of code so long they wrap thrice
|
|
#
?
Aug 4, 2017 06:17
|
|
|
Like seriously, give me a type def operator java
|
|
#
?
Aug 4, 2017 06:17
|
|
|
if you feel the need for typedef you probably need to make a class
|
|
#
?
Aug 4, 2017 06:22
|
|
|
JewKiller 3000 posted:if you feel the need for typedef you probably need to make a class It's more like when you end up with ExtensibleFoo<BarResult, String, BazContainer<Baz>> myExtFoo = ThingReturner<and the rest>() ; I wish I could typedef the type there.
|
|
#
?
Aug 4, 2017 06:36
|
|
|
JewKiller 3000 posted:you should try java, it will teach you how to actually program and get a real legitimate job 
|
|
#
?
Aug 4, 2017 07:06
|
|
|
cis autodrag posted:It's more like when you end up with ExtensibleFoo<BarResult, String, BazContainer<Baz>> myExtFoo = ThingReturner<and the rest>() ; C++'s auto works p good here
|
|
#
?
Aug 4, 2017 07:07
|
|
|
Poopernickel posted:C++'s auto works p good here Java hypothetically had the <> operator but this scenario it doesn't work for. I don't know if I'd want var or auto for this as I'd one the lhs to still carry some meaning, but being able to do like Java code:Would be awesome.
|
|
#
?
Aug 4, 2017 07:18
|
|
|
<T extends HotMess<OfShit>> void doStuff() { T thing = getThing() } You might need to have a parameter or something that actually uses that type, idk. And you're really just punting the problem up to whatever calls this method.
|
|
#
?
Aug 4, 2017 09:54
|
|
|
|
| # ? Apr 25, 2024 16:48 |
|
|
maskenfreiheit posted:as a millennial with limited attention span I can only use interpreted language s like Python kids today never make time for smalltalk
|
|
#
?
Aug 4, 2017 14:18
|
|
