cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end ) | use cffi for calling C | if a pep adds a mere 25-30 [C-API] functions or so, it's a drop in the ocean (cough) - Armin
CrazyPython has joined #pypy
kingsley has joined #pypy
<kingsley> Why would
<kingsley> $ aptitude install pypy3
<kingsley> hang with
<kingsley> "running pypy3 rtupdate hooks for 7.1"
<kingsley> while using ~100% CPU?
<kingsley> It appears to be busy doing...
<kingsley> /usr/bin/pypy3 /usr/bin/pypy3compile /usr/lib/python3/dist-packages
glyph has quit [Remote host closed the connection]
glyph has joined #pypy
CrazyPython has quit [Read error: Connection reset by peer]
speeder39_ has joined #pypy
<kingsley> It finally finished.
<kingsley> I bench marked it on the floating point code at
<kingsley> It was over 2X as fast as python 3.7.3.
<kingsley> Feel free to tell up how fast
<kingsley> $ curl 'https://bpaste.net/raw/peDz' | pypy3
<kingsley> is on your hardware.
<kingsley> Mine?
<kingsley> 46 megaflops
[Arfrever] has quit [Quit: leaving]
[Arfrever] has joined #pypy
glyph has quit [Quit: End of line.]
glyph has joined #pypy
zmt01 has joined #pypy
zmt00 has quit [Ping timeout: 245 seconds]
lritter has quit [Ping timeout: 240 seconds]
lritter has joined #pypy
dddddd has quit [Remote host closed the connection]
_whitelogger has joined #pypy
_whitelogger has joined #pypy
speeder39_ has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #pypy
jvesely has quit [Quit: jvesely]
jvesely has joined #pypy
_whitelogger has joined #pypy
speeder39_ has joined #pypy
_whitelogger has joined #pypy
ajlawrence has joined #pypy
Guest60861 has quit [Ping timeout: 246 seconds]
ekaologik has joined #pypy
lritter has quit [Ping timeout: 268 seconds]
speeder39_ has quit [Quit: Connection closed for inactivity]
jvesely has quit [Quit: jvesely]
<tumbleweed> fijal: looks like that aarch64 jit issue is GC-related, the assertion is: RPyAssert(1, "'contains_weakptr' specified for a large object");
<arigato> no, that really means "some random piece of code corrupted data"
<arigato> or more likely, the traceback you're getting is bogus, because RPyAssert(1) cannot fail
xorAxAx has quit [Quit: Idle timeout reached: 172800s]
Ai9zO5AP has joined #pypy
jacob22_ has joined #pypy
jacob22 has quit [Read error: Connection reset by peer]
bendlas has quit [Write error: Connection reset by peer]
edd[m] has quit [Read error: Connection reset by peer]
agates[m] has quit [Write error: Connection reset by peer]
jacob22_ has quit [Read error: Connection reset by peer]
jacob22 has joined #pypy
bendlas has joined #pypy
tazle has quit [Ping timeout: 252 seconds]
tazle has joined #pypy
ekaologik has quit [Read error: Connection reset by peer]
mvantellingen has quit [Ping timeout: 246 seconds]
CrazyPython has joined #pypy
edd[m] has joined #pypy
agates[m] has joined #pypy
dddddd has joined #pypy
CrazyPython has quit [Remote host closed the connection]
CrazyPython has joined #pypy
tazle has quit [Ping timeout: 246 seconds]
tazle has joined #pypy
CrazyPython has quit [Remote host closed the connection]
gracinet has joined #pypy
<mattip> so do we release without aarch64 py3.6?
jvesely has joined #pypy
CrazyPython has joined #pypy
<arigato> I guess so
<arigato> fijal is unlikely to have time and I'm not home until next weekend
<kenaan> Yannick_Jadoul py3.7 02a99370f6b3 /pypy/: Implemented __mro_entries__ from PEP 560
<kenaan> Yannick_Jadoul py3.7 01b4c2af5928 /pypy/: Fixed annotation errors related to fixed-sized lists, and added a test checking __orig_bases__ is only set w...
<kenaan> Yannick_Jadoul py3.7 0c9e1e22eaad /pypy/: merge upstream py3.7
CrazyPython has quit [Read error: Connection reset by peer]
<arigato> but
<arigato> the first step would be to have intructions on how to reproduce
<arigato> tumbleweed: sorry if you did send such instructions, it seems we can't access the irc logs from Oct 3rd or earlier
<ronan> arigato: there are logs going farther back at https://freenode.irclog.whitequark.org/pypy/
<arigato> thanks
<kenaan> rlamy py3.7 dff15039dbab /pypy/interpreter/: Merged in Yannick_Jadoul/pypy/py3.7-pep562 (pull request #670) Implementation of PEP 562, adding __getattr__ and __d...
<kenaan> Yannick_Jadoul py3.7-pep562 bb000a053f9f /pypy/interpreter/: Implementation of PEP 562, adding __getattr__ and __dir__ for modules
antocuni has joined #pypy
<bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/7730 [ronan: force build, py3.7]
<mattip> arigato: there is also the recurring failure to translate with
<mattip> Exception: 'no_collect' function can trigger collection: <function _siphash24 at 0x0000aaaad786dbc8>
<simpson> Does anybody have any reading on the design of object layers? I've read PyPy's, but I'm still chewing through the general non-Python-specific concepts.
<arigato> I have no clue about this one
<arigato> I'm fine if you ignore aarch64 on py3.6
<arigato> we can always add it later
<Alex_Gaynor> simpson: you can go read topaz's if you want to see a ruby from a few years ago
<mattip> if it is fixable, it would be better to delay the release
<arigato> unsure I agree, but that will have to wait next weekend from my side
<simpson> Alex_Gaynor: Oh yeah, that's another codebase I've read some of. Good idea, thanks.
<mattip> hmm. Anyone else have an opinion?
<simpson> mattip: From the downstream distro side, maintainers are quite okay with marking packages as broken/not-building/different-flags on aarch64.
<mattip> simpson how do we let them know? I think once we tag the release, the downstream maintainers just pull that and build, no?
<mattip> maybe a question for tumbleweed ^^^
<simpson> mattip: If it fails hard enough, then yeah, maintainers will figure it out. It's still good to let people know what doesn't work.
<arigato> and there's always the possibility that the two problems appear on aarch64 but is actually general, it just didn't show up so far
<cfbolz> mattip: i am +0 to not shipping aarch64 3.6 and releasing soonish
<cfbolz> We should document it in the release notes
<mattip> arigato: agreed, that makes sense. Something about alignment checks that are creating collectable objects?
<mattip> (not clear to me what the difference is between a function that triggers collection and one that doesn't, so ignore if that makes no sense)
<arigato> not impossible
<arigato> a function that triggers collection is just a function that gets a "malloc" operation in the end
<arigato> in this case, the function returns a 4-tuple, but it should be inlined---but in this case it isn't
<mattip> looking at that function, there is (misaligned_is_fine or (rffi.cast(lltype.Signed, addr_in) & 7) == 0)
<mattip> but how can that cause a malloc?
<mattip> ok, so agreeing with cfbolz. Let's see if I can formulate a "no aarch64 python3" messasge for the release note that does not convey "aarch64 is not ready"
<mattip> something about "it works on the hardware we could test on" or so
BPL has joined #pypy
antocuni has quit [Ping timeout: 240 seconds]
abrown has quit [Remote host closed the connection]
abrown_ has joined #pypy
oberstet_ has joined #pypy
oberstet has quit [Ping timeout: 245 seconds]
<bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/7730 [ronan: force build, py3.7]
lritter has joined #pypy
xorAxAx has joined #pypy
<kenaan> mattip _siphash24-collecting 1fdf07c8f10e /rpython/rlib/rsiphash.py: try to align the starts differently and pass translation on aarch64
<bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/108 [mattip: force build, _siphash24-collecting]
<bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/108 [mattip: force build, _siphash24-collecting]
<bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/109 [mattip: force build, _siphash24-collecting]
<mattip> \/tmp was using 11GB on the aarch64 buildslave
<tumbleweed> arigato: I can trivially reproduce by trying to build the cffi modules (or do anything non-trivial)
CrazyPython has joined #pypy
<mattip> tumbleweed: this is a PyPy you built, correct? Does it happen with the rc?
<tumbleweed> yes, I built it
<mattip> You can rebuild the cffi based modules locally to get a functioning _ssl
<tumbleweed> let me try your pre-built one
<tumbleweed> seems to work
<tumbleweed> so, something abuot the toolchain
<tumbleweed> I verified it on HEAD, with python ../../rpython/bin/rpython -Ojit targetpypystandalone.py so, nothing about my patches/config
<tumbleweed> it's this easy to trigger on my build: https://paste.debian.net/1104927/
* tumbleweed starts a build with gcc 8 (rather than 9)
<mattip> what python did you use?
<mattip> to translate with
<tumbleweed> cpython
* tumbleweed should file an issue with what he knows...
<mattip> which version of cpython?
<tumbleweed> this is debian unstable, so 2.7.16
CrazyPython has quit [Read error: Connection reset by peer]
<mattip> I can try to reproduce on our buildslave later on
<mattip> we use PyPy2 to translate usually
<tumbleweed> I would usually, too
<tumbleweed> but bootstrapping arm64, here
<tumbleweed> https://bitbucket.org/pypy/pypy/issues/3086/arm64-jit-lots-of-crashes has my build-log with all toolchain versions
<tumbleweed> bah, no toolchain versions because that was a manual build
CrazyPython has joined #pypy
CrazyPython has quit [Read error: Connection reset by peer]
gracinet has quit [Quit: Leaving.]
gracinet has joined #pypy
gracinet has quit [Quit: Leaving.]
<bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/109 [mattip: force build, _siphash24-collecting]
speeder39_ has joined #pypy
oberstet_ has quit [Ping timeout: 240 seconds]
BPL has quit [Quit: Leaving]
oberstet has joined #pypy