<whitequark>
I wrote a hindley-milner-style inferencer for py2llvm during the last month
<whitequark>
from a quick look it does not appear that hermetic would have made development of this faster
<rjo_>
not in current artiq we don't (non-inlined functions that is)
<rjo_>
is that inferencer in the artiq branch or in pythonparser?
<rjo_>
then, once new-py2llvm is done, we will have functions in artiq. that is earlier than planned...
rjo_ has quit [Quit: leaving]
<whitequark>
new-py2llvm
<whitequark>
er
<whitequark>
I mean it's in artiq.
<whitequark>
pythonparser concerns itself only with python, and python is untyped.
<whitequark>
rjo: not just functions, closures too
key2 has joined #m-labs
bentley` has joined #m-labs
<sb0>
whitequark, how do you support closures without a heap?
<sb0>
(and gc)
<sb0>
ok, runtime no longer crashes after adding a self powered usb hub
<whitequark>
sb0: a closure is just a mutable record packed inside a function
<whitequark>
i.e. in the exact same way as any other mutable thing, without a heap
<sb0>
but the scope of the data that form that mutable record is only the function that creates the closure
<sb0>
ewww it does
<sb0>
dammit
sb0 has quit [Remote host closed the connection]
sb0 has joined #m-labs
<sb0>
the shittiness of the baytrail emmc driver would be a perfect test for journaled filesystems. unfortunately, those are buggy too and you still get data corruption.
<whitequark>
sb0: sure
<whitequark>
this is the exact same case as, say: a=[]; b=[a]
sb0 has quit [Read error: Connection reset by peer]
sb0 has joined #m-labs
<GitHub116>
[artiq] sbourdeauducq pushed 4 new commits to master: http://git.io/vt02G
<GitHub116>
artiq/master e6a4c2f Sebastien Bourdeauducq: dds: make it easier to specify phase
<GitHub116>
artiq/master c71fe29 Sebastien Bourdeauducq: simplify unit system and use floats by default