DuToitSpies has quit [Quit: Konversation terminated!]
DuToitSpies has joined #hpy
DuToitSpies has quit [Ping timeout: 248 seconds]
DuToitSpies has joined #hpy
DuToitSpies has quit [Ping timeout: 272 seconds]
DuToitSpies has joined #hpy
Techcable has quit [Ping timeout: 256 seconds]
Techcable has joined #hpy
steve_s has quit []
steve_s has joined #hpy
<Hodgestar>
cfbolz: Do you perhaps have a standard set of starting references on mondern dynamic language implementations that you could point DuToit at?
<Hodgestar>
fangerer: And the same question but perhaps specifically for Graal?
<fangerer>
If DuToit wants to learn about GraalPython's implementation, then yes. Is that what you mean?
GianlucaRizzo has joined #hpy
<Hodgestar>
fangerer: Yep!
<GianlucaRizzo>
Hey everyone, I'm Gianluca Rizzo, a computer science honours student from UCT. I'm working with Du Toit on HPy for my honours project. I'm looking forward to working on it, it seems very interesting!
* Hodgestar
waves.
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
dutoi has joined #hpy
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
DuToitSpies has quit [Ping timeout: 246 seconds]
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
dutoi has quit [Ping timeout: 250 seconds]
DuToit has quit [Ping timeout: 250 seconds]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
GianlucaRizzo has joined #hpy
<cfbolz>
Hodgestar: any specific direction? Roughly how many papers?
<cfbolz>
There is no good annotated bibliography. One approach is to follow the references in these slides: https://t.co/fuAzCHxy4u
<Hodgestar>
cfbolz: DuToit and Gianluca are doing literature reviews before beginning their investigating-hpy-performance projects. I was thinking something like one or two review papers that link to lots of other papers. DuToit and Gianluca can ask for more guidance later if they need it.
<Hodgestar>
cfbolz: Perfect!
<cfbolz>
There's a somewhat dated but still very interesting survey paper by John Aycock, 'A Brief History of just in time' is the title i think
DuToit has quit [Ping timeout: 272 seconds]
GianlucaRizzo has quit [Remote host closed the connection]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
GianlucaRizzo has joined #hpy
<antocuni>
back from PyCon DE and reading the backlog of messages, so I am probably late for replying
<cfbolz>
antocuni: how was pyconde?
<antocuni>
but about the debug mode, it depends whether we want to allow per-module debug mode or not. If think we should, unless there are very good reasons to avoid it
<antocuni>
for example, if I am debugging my own module, I don't want to turn on debugging for other unrelated modules
<antocuni>
(or maybe there are cases where I want, but that should not be enforced, IMHO)
<antocuni>
so we also need a way to specify for which modules we want to enable the debug mode. At a first approximation, I think an ENV var can be fin
<antocuni>
then if we want to be advanced we can also provide command line arguments, .ini files, and a way to specify them from python
<antocuni>
cfbolz: very very good
<antocuni>
it was a good feeling to be back to in-person conferences
DuToit has joined #hpy
<Hodgestar>
Yay for having a good conference!
<Hodgestar>
Agreed about debug mode being per-module if at all possible. Would HPY_DEBUG=numpy imply debug mode for all modules inside the numpy package?
<cfbolz>
antocuni: awesome
<antocuni>
Hodgestar: good question
<cfbolz>
What were your impressions on the web assembly stuff? Does it have actual use cases or mostly a tech demo?
<antocuni>
my gut feeling is that you should list the exact module name which you are loading, in this case it would be HPY_DEBUG=numpy.core._multiarray (or whatever it's called nowadays)
<antocuni>
but we can also be more advanced and say e.g. HPY_DEUBG=numpy.* maybe?
<antocuni>
cfbolz: I think that for now it's just a demo
<antocuni>
the idea is that CPython could/should support webassembly as an official target
<antocuni>
so that e.g. pyiodide can concentrate with the actual integration with the browser, instead of spending time at fixing CPython to work
<cfbolz>
antocuni: ah richtig
<cfbolz>
Right
<antocuni>
I also think it's still kind of slow, but Christian says that lots of optimizations are possible
<cfbolz>
The only use case I understand is zero install education
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
<antocuni>
there is also jupyterlite, which if I understand correctly it's like jupyter notebook but runs entirely client side
<cfbolz>
antocuni: yeah, that's kind of what I meant
GianlucaRizzo has quit [Remote host closed the connection]
GianlucaRizzo has joined #hpy
DuToit has quit [Read error: Connection reset by peer]
GianlucaRizzo has quit [Ping timeout: 240 seconds]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 256 seconds]
FFY00_ has joined #hpy
FFY00 has quit [Ping timeout: 260 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
GianlucaRizzo has joined #hpy
DuToit has quit [Ping timeout: 272 seconds]
DuToit has joined #hpy
GianlucaRizzo has quit [Remote host closed the connection]
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 248 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 248 seconds]
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 248 seconds]
Techcable has quit [Remote host closed the connection]
Techcable has joined #hpy
DuToit has quit [Ping timeout: 248 seconds]
DuToit has joined #hpy
DuToit has quit [Read error: Connection reset by peer]
DuToit has joined #hpy
GianlucaRizzo has joined #hpy
GianlucaRizzo has quit [Ping timeout: 250 seconds]
<Hodgestar>
Zero install (or really install-so-easy-you-can-miss-it) is pretty compelling in non-educational contexts too (e.g. in a lot of businesses getting non-devs or even non-scientific Python devs to install jupyter-notebook plus tools is large mountain to climb).
<Hodgestar>
But running in a browser does come with many downsides and making building and install packages easier solves some of the installation problem from a different direction.
<Hodgestar>
I think it's good that people are doing some strange things which might be really useful if they pan out. :)