Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
Repo info
  • Mar 03 2017 15:52
    @dom96 banned @Octopoda7
  • Feb 12 2017 23:57
    @dom96 banned @zzz125
  • Dec 23 2016 19:43
    @dom96 banned @Izrab
@livcd yes
From IRC (bridge bot)
<Zevv> wow lets make an m4 compatible source filter
Alexander Ivanov
+ term rewriting macros
From IRC (bridge bot)
<livcd> @gogolxdong: Do you know xcgui ?
<PMunch> livcd, hard to find any english docs on it it seems
what's that?
From IRC (bridge bot)
<livcd> some chinese gui lib
<PMunch> Windows only by the looks of it
yeah, it's Windows only.
From IRC (bridge bot)
<PMunch> Hmm, what are the current differences between a normal build and -d:release?
<PMunch> My program fails when I compile with -d:release on 0.20.2, but works fine without
<Zevv> "fails"?
<PMunch> It's doing some decryption, which comes out garbled on a release build
From IRC (bridge bot)
<leorize> I think the only diff are: stacktrace off, -O3 optimization on
<leorize> I think -O3 is messing up with your code, try passing -d:release -t:'-O0' and see if it's working
<PMunch> Nope, still messed up
<PMunch> Was that supposed to be -t
<leorize> it's --passC, I think the sort form is -t
<leorize> try --opt:none then
<PMunch> Ah, it's definitely -O3 that does it
<PMunch> But passing -O0 to a release build adds both flags, and the GCC compiler chooses O3
<PMunch> Just passing O3 without release messes up as well
<PMunch> Some for O1
From IRC (bridge bot)
<Zevv> So O3 messes up your nim-generated code?
<PMunch> Hmm, strange, this used to work..
<PMunch> Zevv, I don't think O3 does anything to the code does it?
<Zevv> well, it can break bad code
<Zevv> or fix it, depending on how you look at it
<Zevv> it typically changes UB
<Zevv> but nim generated C code should never show UB, right
<leorize> well gcc optimization make code fails faster :⁠P
<Zevv> that much is true :⁠)
<leorize> I'm not entirely sure if -O3 is that safe
<Zevv> well, it is supposed to be
<leorize> libgcrypt has special override for certain part of it to disable gcc optimization
<leorize> PMunch: if you have clang with sanitizer, try -t:'-fsanitizer=ubsan' -l:'-fsanitizer=ubsan'
<PMunch> Only got GCC on this machine
From IRC (bridge bot)
<Zevv> Does O2 work for you?
<PMunch> Nope
<Zevv> that is very bad
<PMunch> Only O0 works
<leorize> can you share the code?
<PMunch> O[1|2|3] fails
<PMunch> Sure, it's my libkeepass repo on GitHub
<leorize> if -O1 fails then the code is already broken :⁠P
<Zevv> euh but that's a python module?
<Zevv> are you sure you're on the right irc channel? :⁠)
<PMunch> It's a Nim module