antocuni changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | "PyPy: the Gradual Reduction of Magic (tm)"
danieljabailey has quit [Quit: ZNC 1.6.4+deb1 - http://znc.in]
<simpson>
Hi! I have an irate and ranty community member who is wondering what license the output of an RPython translation is under, in the case where they use rpython.rlib.
<_stian>
@simpson "all software and documentation in the 'rpython', 'pypy', 'ctype_configure', 'dotviewer', 'demo', 'lib_pypy', 'py', and '_pytest' directories is licensed as follows: "
<arigato>
I don't think the output of any compiler is protected under a specific license different from the input, is it?
<cfbolz>
arigato: it can be. we link against rlib/gc code
<cfbolz>
if that code is gpl, there might be problems
<arigato>
ah
<arigato>
ok
<cfbolz>
which I think is why the question specifically mentioned rlib
<cfbolz>
but yes, as far as I understand it MIT is permissive enough
<arigato>
indeed, and we could have decided our GC is 100
<arigato>
100$/program
<antocuni>
I think that gcc has a special "Runtime Exception" clause specifically to avoid forcing the output program to be gpl
<simpson>
arigato: Golang and Rust both have reference compilers which compile parts of their stdlib into binaries, preserving the license.
<simpson>
And yes, GCC and LLVM have exemptions for their static pieces.
<simpson>
The user in question is IMO unreasonable; they want to allow their downstreams to totally omit attribution, which seems like a bad choice. OTOH it's good to know what's allowed and what's not.
<xorAxAx>
antocuni: indeed
jacob22 has quit [Ping timeout: 240 seconds]
jacob22 has joined #pypy
<kenaan_>
cfbolz default b95f1240ad90 /pypy/interpreter/pyparser/: fix issue #2701 allow the sequences future-import, docstring, future-import for CPython bug-compatibility
antocuni has quit [Ping timeout: 268 seconds]
songww has quit [Remote host closed the connection]
_stian_ has joined #pypy
songww has joined #pypy
tav has quit [Quit: tav]
<_stian_>
Updated pull request with the benchmarks.
<fijal>
arigato: alive?
<arigato>
zzkind of
<fijal>
hehe
* fijal
slept 15h
<fijal>
and I STILL had to set up an alarm clock
songww has quit [Quit: songww]
lritter_ has quit [Ping timeout: 248 seconds]
Rhy0lite has joined #pypy
dddddd has joined #pypy
<kenaan_>
stian math-improvements 3f4aca709e49 /rpython/rlib/: Remove invert logic from rqshift (it is only used with positive numbers)
<fijal>
25% in GC is a lot right?
fryguybob has quit [Remote host closed the connection]
<mva>
is it any way to definitely say them that "yes, it is PaX system" and avoid that check?
<fijal>
what's PaX?
<kenaan_>
rlamy default 3b8c612bb506 /pypy/module/cpyext/test/test_methodobject.py: Test an obscure difference between C-defined and Python-defined functions
marky1991 has quit [Read error: Connection reset by peer]
marky1991 has joined #pypy
<mva>
fijal: kernel patchset, that kinda "protects" from 0-day exploits
<mva>
(when DoS is acceptable and better than, say successful buffer overflow attack)
marky1991 has quit [Ping timeout: 248 seconds]
<nanonyme>
njs, incorrect. Hope you aren't shipping any programs with that assumption
<nanonyme>
njs, Microsoft is *not* shipping the C Runtime with the OS. They're shipping a loader library called UCRT instead. It doesn't do anything unless you actually ship C Runtime
<nanonyme>
The only thing that's changed is Windows now has *part* of C Runtime included that you're *not* supposed to ship so situation is *worse* than before
raynold has joined #pypy
<nanonyme>
Staticly linking has the benefit that you're then not at Microsoft's mercy for shipping the MSU that gives UCRT nor are you forced to push content down Windows Update system yourself
<tos9>
Anyone here on OSX mind double checking something
<tos9>
Actually I should test on CPython to confirm it's only PyPy... I just assumed it was otherwise everyone'd be complaining.
Rhy0lite has quit [Quit: Leaving]
<tos9>
Yeah works fine on CPython.
<Alex_Gaynor>
how does it error out?
<tos9>
Alex_Gaynor: (I just moved to #cryptography-dev, might be better in there?)
<tos9>
Alex_Gaynor: But, with one of the opaque OpenSSL errors with no detail
<Alex_Gaynor>
sure, we can go there
marky1991 has joined #pypy
drolando has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
drolando has joined #pypy
exarkun has quit [Ping timeout: 248 seconds]
tbodt has joined #pypy
exarkun has joined #pypy
drolando has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
drolando has joined #pypy
marky1991 has quit [Ping timeout: 248 seconds]
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
tav has joined #pypy
adamholmberg has quit [Remote host closed the connection]
adamholmberg has joined #pypy
adamholm_ has joined #pypy
adamholmberg has quit [Ping timeout: 240 seconds]
<kenaan_>
rlamy default a305590465d6 /: Implement cpyext.is_cpyext_function() inspect.isbuiltin() now returns True for functions implemented in C, like on...
<nanonyme>
Is that good or bad?
<ronan>
that's one less obscure difference with CPython
<ronan>
though I wonder if this breaks any people's workarounds...
<nanonyme>
Does it now return True for all third-party CFFI modules as well?
<ronan>
oh, good point, I didn't check
<ronan>
AFAICT, the answer is no
<nanonyme>
Okay. I was just a bit concerned after reading before you guys had ssl through CFFI module
adamholm_ has quit [Remote host closed the connection]
<mattip>
this started failing "np.array(['100']).astype(int)"
<mattip>
since on cpython PyStringArrType_Type.tp_as_number.tp_int is NULL but we fill it from gentype_int
exarkun has quit [Ping timeout: 240 seconds]
exarkun has joined #pypy
<mattip>
it feels like I keep fixing this and it keeps breaking somehow
tormoz has quit [Quit: No Ping reply in 180 seconds.]
tormoz has joined #pypy
yuyichao has joined #pypy
abvi[m] has joined #pypy
yuvipanda has joined #pypy
bendlas has joined #pypy
<kenaan_>
arigo default 27b914ed1ea1 /rpython/config/support.py: Merged in tpruzina/pypy (pull request #581) fix detect_pax behavior on linux if procfs is mounted with hidepid>=1 ...
<kenaan_>
Tomáš Pružina default 368d2eef1229 /rpython/config/support.py: fix detect_pax behavior on linux where procfs is mounted with hidepid>=1 PID1 (init) isn't observable on s...