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)"
<mssun>
Hey guys, do you know how to imporove the building time of pypy? I'm using a E7 sever. The compilation time is about 1 hours.
<mssun>
The only thing I can do right now is to set the `--make-jobs` option. But this can only improve the last part of building process - compilation C intermediates.
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
dddddd has quit [Remote host closed the connection]
mac10dieplenty has joined #pypy
Hotpot33 has quit [Ping timeout: 240 seconds]
<mac10dieplenty>
the battle continues for a free market in 2018!!!!
<mac10dieplenty>
ladies and gents, I give to you the largest music exchange on the planet
<mac10dieplenty>
a free exchange with no middleman
<mac10dieplenty>
I've just returned from the future
<mac10dieplenty>
2023
<mac10dieplenty>
and boy I can tell you this shit is hot
<mac10dieplenty>
the largest music exchange on the planet and your finding out about it early on
<mac10dieplenty>
here and now
<mac10dieplenty>
a great fortune unto you
<mac10dieplenty>
I give to you
<mac10dieplenty>
VOISE
<mac10dieplenty>
the south korea turmoil in regards to authorities shutting down crypto exchanges has caused a huge correction to happen for all cryptocurrencies on the market
<mac10dieplenty>
now is the time to strike with everything you've got as everything is on sell at a discount!
<mac10dieplenty>
livecoin.net - best service, intuitive interface, trustworthy, secure
<ethanhs>
Hi, I was looking at GSOC 2017 project list and noticed explicit typing was one of the project
<ethanhs>
would PEP 484 based typing syntax be a possible solution?
<LarstiQ>
ethanhs: from what I recall, that's the wrong kind of information to help the jit
* LarstiQ
would have to search the logs
<simpson>
ethanhs: Maybe, but we'd have to backport the functionality into the RPython toolchain at a deep level, since the surface import is Python 2.7.
<simpson>
More likely we'd want to make something like enforceargs but better.
<ethanhs>
simpson: that is a fair point
<ethanhs>
is there a roadmap to drop 2.7 support?
<cfbolz>
No
<simpson>
It's not a polyglot codebase. It's *only* 2.7. And there's no plans to port to Python 3.
<ethanhs>
Is there a desire to?
<cfbolz>
No. Too much work without clear benefits
<ethanhs>
I see
<ethanhs>
What about the skill gap that will happen?
<simpson>
Python 2 isn't dying, so people will still know it. It's fine.
<simpson>
Also, really one needs to learn RPython, not just Python 2.
<ethanhs>
Python 2 will EOL in 2020, more importantly, few if any institutions are teaching Python 2 anymore
<ethanhs>
(to my knowledge)
<ethanhs>
That is a fair point about rpython however
<simpson>
Don't worry about that EOL; it only applies to CPython. We'll still have a PyPy 2.7 which can build itself.
<ethanhs>
sure, but as the reference implementation CPython has a big influence on the ecosystem
<ethanhs>
Anyway, it seems a decorator approach may indeed be better suited for this project
<ethanhs>
simpson: so then the issue is you want something like Dict from typing to define container types and more complex types?
<simpson>
I think that adding containers to what's currently supported would be a reasonable GSoC thing. But I'm not mentoring, I think.
<ethanhs>
Sure, I'm mostly interested in what that project would look like, which doesn't absolutely need to come from a mentor :)
<ethanhs>
At least at a higher level
<simpson>
Probably a mini-parser for the type mini-language which builds some type objects, and then add logic for the type objects into enforceargs.
<ethanhs>
Interesting, I think I will start sketching that out see what it could look like.
<ethanhs>
Ideally to me it sounds like CPython 2.7 compatibility is desired, so this wouldn't be able to introduce new syntax?
<simpson>
I imagine that it depends on what you want to introduce.
<ethanhs>
Nothing that would be backwards incompatible of course, but perhaps something that CPython wouldn't have. I thought being able to translate PyPy on CPython was important.
<simpson>
It's pretty important, but maybe we'd be willing to trade it for something sufficiently nice. It's more likely though that we'd use some sort of metaprogramming system.
<simpson>
But of course RPython already has lots of metaprogramming tools, so whatever we use would probably be purpose-built just for us.
<ethanhs>
Sure
<ethanhs>
I mean to me the 2.7 suggested syntax for PEP 484 is nice because it is backwards compatible and straightforward. But I'll be looking at other options as well