Tuesday, August 9, 2016

PyPy gets funding from Mozilla for Python 3.5 support

"Python 2.x versus Python 3.x": this is by now an old question. In the eyes of some people Python 2 is here to stay, and in the eyes of others Python has long been 3 only.

PyPy's own position is that PyPy will support Python 2.7 forever---the RPython language in which PyPy is written is a subset of 2.7, and we have no plan to upgrade that. But at the same time, we want to support 3.x. This is particularly true now: a relatively recent development is that Python 3.5 seems to attract more and more people. The "switch" to Python 3.x might be starting to happen.

Correspondingly, PyPy has been searching for a while for a way to support a larger-scale development effort. The goal is to support not just any old version of Python 3.x, but Python 3.5, as this seems to be the version that people are switching to. PyPy is close to supporting all of Python 3.3 now; but the list of what is new in Python 3.4 and 3.5 is far, far longer than anyone imagines. The long-term goal is also to get a version of "PyPy3" that is as good as "PyPy2" is, including its performance and its cpyext layer (CPython C API interoperability), for example.

So, the end result: Mozilla recently decided to award $200,000 to Baroque Software to work on PyPy as part of its Mozilla Open Source Support (MOSS) initiative. This money will be used to implement the Python 3.5 features in PyPy. Within the next year, we plan to use the money to pay four core PyPy developers half-time to work on the missing features and on some of the big performance and cpyext issues. This should speed up the progress of catching up with Python 3.x significantly. We are extremely thankful to Mozilla for supporting us in this way, and will keep you updated on the progress via this blog.

24 comments:

Dave said...

Great to hear of this development. I'm one of those "Python has long been 3 only" developers, but have had an eye on PyPy for a long time and even donated several times. Planning to switch to PyPy when 3.5 support lands.

Unknown said...

glad to hear that.

To me, the time to switch to py3 depends upon the maturity of pypy 3.

I have used pypy 2 for a while in production, so far so good.

Alessandro said...

Great news!

I'm one of those "Python 3 only". The switch was terrible for the community, but Python 3 is superior than 2 in my opinion.

RPython 3 would be great to, but it's propably complete inviable.

Ronan said...

Fantastic news! Thanks for your work, and thanks Mozilla for their support :)

Shen said...

Awesome !

Unknown said...

Is there any chance optional typing information will be used to help the JIT?

Anonymous said...

200,000 sounds like a lot of money but it is two developers for a year at less than Silicon Valley wages.

Anonymous said...

I think you are overpaying yourselves. But hey, it's your money.

Anonymous said...

2.7 forever!

cclauss said...

This is huge news! Corporate sponsorship of open source projects is a beautiful thing for us all. Congrats to the PyPy team. You really deserve this kind of support for all of your hard work and perseverance over the years.

Given that Python 3.6 will be going beta next month, perhaps that should be your target instead of 3.5 but you know your craft better than I do.

Those of you who would be interested to pay Mozilla back for this investment might want to help port the following 9 Mozilla projects to Python 3:
* mozrunner, moznetwork, mozdevice, mozprocess, mozprofile, mozfile, mozinfo, mozlog, mozcrash

These nine Python 2 projects are all in the Top 150 PyPI downloads of all time and each of them has been downloaded with pip more than 5 million times. Currently 92% of the Top 200 PyPI packages are Python3 compatible. Converting these 9 Mozbase modules to Python 3 would bump that number up to 96.5%. It would also probably push us over the line where 50% of the Top 4,000 PyPI packages are Python 3 compatible. This kind of momentum would be welcome news as the Python community continues our move to Python 3.

Armin Rigo said...

Why Python 3.5 instead of 3.6? That's because 3.5 is the version that attracts people. Python 3.6 will only be out of beta in december and my own wild guess is that it won't immediately attract all the 2.7 people, who grew accustomed to sticking to a known unchanging version. So the deal with Mozilla is to get a PyPy 3.5 working as well as possible; it is better than getting a PyPy 3.6 that (like current versions of PyPy 3) has downsides in completeness and performance.

Unknown said...

Great news! I Can't wait to the moment we get an stable Python3 PyPy! Congratulations!

Unknown said...

Great news! I Can't wait to the moment we get an stable Python3 PyPy! Congratulations!

guillem said...

TL;DR. Python 3.5 is a "good enough" and seems a future-proof language.

I was in charge of deciding which version of Python to use at my job. We started the development of a framework supporting Python 2.7 and Python >=3.4, but we recently switched to a Python 3 only development. The whole python 2 vs 3 thing was quite confusing to the operations department and developers that are not proficient with Python.

There was a quite thorough assessment of the features, and we decided to stick to Python 3.5, at least for the next decade or so. On the Python 3 side, one of the reasons was that the async/await syntax allows junior developers to understand (more or less) asynchronous programming, while coroutines+decorators are quite a mess. We still have some Red Hat instances that use Python 3.4, but as soon as we get rid of them, everything will be Python 3.5.

touilleMan said...

Awesome news ! Long live Pypy ;-)

Anonymous said...

Super great news!

Anonymous said...

This is amazing news! I use Python3 extensively in private research projects. Unfortunately, I have been in the position of choosing between Python3 and having a high-performance interpreter implementation. Testing the PyPy 3.3.5 alpha shows all-around disappointing performance in our string-manipulation-heavy projects making intense use of unicode, so I can only recommend our team to stay with CPython for both performance and compatibility.

I am very excited to hear that PyPy3 will be getting more of the specific attention it deserves, with Python 3.5 support to boot!

PvdE said...

Good news, and you definitely deserve it. But I guess this will take virtually all of the PyPy team's resources for the next one-two years, what does this means for other in-progress innovations, in particular STM-GC? I donated but it looks like the money is not being spent.

If continuing the improvements to CPython support means PyPy will become an option for more programs, that would be great. The (few) red lines in http://packages.pypy.org/ and lower performance for others are still blockers for many users.

Armin Rigo said...

@PvdE: I'll admit that the STM-GC is not actively going anywhere right now. STM is hard, unsurprizingly. There is still a bit being done by Remi (working at an academic institution where he doesn't need the money). For me, I am part of the Python 3.5 team, but I might also come back to STM-GC soon. We expect not all our resources to be consumed by Python 3.5. In fact, the money covers four half-time jobs (and flexibility allows someone to do less than half-time while someone else does more).

Anonymous said...

This is great news.

A developer's decision to target Python 3 over Python 2 in their projects is more fundamental than deciding which interpreter to use. People use Python 3 because it's future-proof, to take advantage of its new features and to do their bit in driving the Python ecosystem forward. For me and I imagine many others, being curious about PyPy hasn't been enough to override all those factors.

I think there's a huge untapped audience out there waiting for first-class support for modern Python in PyPy before giving it a shot. I hope to see a big bump in PyPy adoption with this move, and a corresponding bump in funding and support for PyPy's development.

Thanks for your fantastic work!

JP said...

Great news! Will this include making numpy work with Python3 pypy? That's the main thing preventing me from evaluating pypy for my Python3-only OpenGL application.

mattip said...

@JP cpyext compatibility is one of the milestones, and we currently pass over 99% of the upstream NumPy test suite using PyPy 2.7, so it all should Just Work

Anonymous said...

Getting PyPy3 to 3.5 status is a good start considering that the current LTS version of Ubuntu (16.04) has 3.5 and that is going to be supported for a while.

Anonymous said...

Great news! Pypy's lack of Python 3 support is the biggest reason I haven't switched yet. It technically supports most of 3.3 already, but since Pypy3 is slower than CPython, it may as well not exist. Hopefully you'll also work out the performance issues in Pypy 3 as well.