cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | the secret reason for us trying to get PyPy users: to test the JIT well enough that we're somewhat confident about it
drolando has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
drolando has quit [Remote host closed the connection]
drolando has joined #pypy
inad922 has quit [Ping timeout: 264 seconds]
<kenaan>
Julian Berman pypy.org[extradoc] e7d9800ec5ac /source/download.txt: This warning is apparently not too relevant anymore, at least in this spot.
<kenaan>
Julian Berman pypy.org[extradoc] 913450bf91e9 /download.html: Regenerating HTML
<tos9>
OK on to the next thing in the yak stack... numpy appears to be failing at the minute for me on 5.10 (which always seems to be me being dumb, but I'm not seeing what I'm doing wrong quite yet)
<tos9>
Ah, or maybe the build output is just confusing and I do know what I'm doing wrong
<tos9>
The first error is for things like _configtest.c:1:10: fatal error: 'endian.h' file not found
<tos9>
But that appears to just be numpy randomly trying to find the right way to #include that I guess?
<mattip>
tos9: what OS?
<tos9>
OSX
<mattip>
hmm, then no idea
<tos9>
mattip: I am pretty sure I'm missing dependencies (BLAS?) so I am expecting it to fail because of that
<mattip>
no, that should not be a showstopper, but it will be slower
<tos9>
Hm, OK
<tos9>
mattip: It's failing in from what I can tell a weirdly early place ... will paste the output just in case anything looks particularly suspicious
<mattip>
perhaps start with building on CPython, then try with PyPy
<tos9>
(that's from the pypy install... let's see a --no-binary cpython)
<tos9>
ok yeah it failed on cpython with --no-binary
<tos9>
guessing maybe the error about not knowing how to compile fortran might be the important one... let's see what happens after installing gfortran
inad922 has joined #pypy
amaury has quit [Quit: Konversation terminated!]
amaury has joined #pypy
<njs>
tos9: numpy doesn't have any fortran code
<njs>
oh interesting, it's decided it should try to link against Apple's Accelerate library, but failing for some reason
<njs>
maybe it just assumes that it's available on MacOS, and you're missing some important SDK or something?
planrich has quit [Ping timeout: 264 seconds]
<tos9>
njs: Yeah so I'm googling the Accelerate error
<tos9>
but google just seems to suggest that I should be needing XCode
<tos9>
(Which I already have.)
<tos9>
njs: (installing gfortran did change the build output, but it still does fail yeah)
<mattip>
tos9: please document the steps you take and post it somewhere, so the next entrepid adventurer will be able to follow your footsteps
stduolc has joined #pypy
<tos9>
mattip: will do!
amaury has quit [Ping timeout: 246 seconds]
<tos9>
(and probably will post it to a numpy bug to have them document it somewhere around where they say how to compile from source)
<tos9>
first lunch though
illume has joined #pypy
mcdoker181818 has joined #pypy
planrich has joined #pypy
<kenaan>
mattip py3.5 fce0870ca033 /pypy/module/cpyext/test/: add missing files for tests
<kenaan>
mattip default 56c6bd396489 /pypy/module/posix/: raise OSError not IOError (py2.7, linux failing stdlib test)
<dash>
(this may be the greatest python crime i've ever committed)
<cfbolz>
I am clearly not smart enough for you macropy people
jredondo__ has joined #pypy
<dash>
cfbolz: My fear is that in a few months I won't be smart enough for this code either.
jredondo__ has quit [Ping timeout: 240 seconds]
<tos9>
what is macropy and why is macropy
<tos9>
oh geez...
<simpson>
tos9: This is GvR's fault for not giving us do-notation~
<cfbolz>
I think the answer is "lispers gonna lisp"
<tos9>
simpson: you mean it's GvR's fault for giving us Python? FTFY
<tos9>
nah yeah fine yours was fine as is
<dash>
tos9: look upon my works and despair, etc
<dash>
to be fair the author of macropy acknowledges that you can do most things in python without macros
<dash>
but rpython forbids a lot of those things, so here i am
<arigato>
rpython is powerful enough to have its own rpython-specific ways to do most of these things, but if you prefer some macro system you're more used to, that's fine too
<simpson>
arigato: We just want promises.
<dash>
yeah promises is where I started but there wasn't an obviously reasonable way to make the types work out
<arigato>
so, you'd like something like "Promise<T>" in C++/Java/C#, but how to write that in RPython is indeed obscure
<simpson>
Yeah, that'd be okay. I'd also settle for Either as in OCaml/Haskell/F# and then we can do asynchrony by hand.
<arigato>
if you're thinking of Either you're not used enough to non-functional programming :-)
<arigato>
or more precisely, OOP
<simpson>
Nah, nah, it's tried and tested. Deferreds, for example, are very much like Either. They aren't async by themselves.
<simpson>
RPython isn't as object-oriented as I'd like. We can't send messages, for example, which is ultimately what this promise feature is supposed to enable.
<simpson>
But I hear what you're saying. We're not very RPythonic.
planrich has quit [Ping timeout: 256 seconds]
antocuni has joined #pypy
krono has quit [Quit: Connection closed for inactivity]
tayfun26 has quit [Remote host closed the connection]
<cfbolz>
all these things are too higher order for RPython's poor naive type system though
<xorAxAx>
RPython II: Dependant types and effects.
<cfbolz>
I'd rather have "RPython II: proper errors and separate compilation"
inad922 has joined #pypy
oberstet has quit [Ping timeout: 240 seconds]
inad922 has quit [Ping timeout: 240 seconds]
raynold has joined #pypy
inad922 has joined #pypy
inad922 has quit [Ping timeout: 264 seconds]
asmeurer__ has joined #pypy
asmeurer__ has quit [Ping timeout: 264 seconds]
planrich has joined #pypy
illume has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
energizer has joined #pypy
asmeurer has joined #pypy
illume has joined #pypy
mcyprian has joined #pypy
antocuni has quit [Ping timeout: 240 seconds]
mcyprian has quit [Client Quit]
asmeurer has quit [Quit: asmeurer]
jredondo__ has joined #pypy
jredondo__ has quit [Ping timeout: 260 seconds]
tbodt has joined #pypy
<mattip>
antocuni: f428c66de212 exposed a subtle bug, PySequence_ITEM calls space.getitem,
<mattip>
which will prefer tp_as_mapping.mp_subscr over tp_as_sequence.sq_item
<mattip>
both map to __getitem__
lritter has quit [Quit: Leaving]
illume has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
nunatak has joined #pypy
tbodt has joined #pypy
tbodt has quit [Read error: Connection reset by peer]
<mattip>
there is an optimization in buffer_concat, "if (size == 0) { ... return other;}"
marky1991 has quit [Ping timeout: 276 seconds]
illume has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
marky1991 has joined #pypy
marky1991 has quit [Remote host closed the connection]
marky1991 has joined #pypy
marky1991 has quit [Remote host closed the connection]
inhahe_ has joined #pypy
marky1991 has joined #pypy
inhahe__ has quit [Ping timeout: 252 seconds]
inhahe has quit [Ping timeout: 276 seconds]
inhahe has joined #pypy
<kenaan>
mattip default abd1818fcde4 /pypy/module/cpyext/: test, fix for PySequence_ITEM accessing sq_item, not mp_subscr. Both are __getitem__, the second takes precedence
<kenaan>
mattip default d044ce2d7d53 /pypy/: test, fix for cpython2 optimization where "buffer('') + obj is obj"
<mattip>
found the buffer issue when running tests with -A
forgottenone has quit [Quit: Konversation terminated!]