Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Scipy 1.0 released (python.org)
370 points by ngoldbaum on Oct 25, 2017 | hide | past | favorite | 36 comments


Scipy+numpy made it possible for me to mostly stop using the various weird, usually proprietary and often highly specialized data analysis applications that are entrenched in the scientific community, and work with a nice, general purpose programming language instead. It was wonderful!

This is close to a decade ago, it kind of blows my mind that only now scipy 1.0 is released. Not that it matters, I'm still a user, thanks team scipy!


Having spent many years writing arcane integration layers to sit between user-facing services and the weird, usually proprietary and often highly specialized data analysis applications that were being used to do data processing, the ability to use the same language for everything saves years of time, effort and complexity.


Agreed, many congrats due to the Scipy team! Their library was one of several that helped me get away from MATLAB post graduate school.


It's really momentous that at LAST there is a Windows binary distribution available on PyPI. `pip install scipy numpy matplotlib` finally just works when you install Python on Windows.


For years I used Gohlkes installers/wheels[0] to install a great deal of libs on windows easily, SciPy amongst them. For a few packages, it still is the best alternative.

[0] http://www.lfd.uci.edu/~gohlke/pythonlibs/


It's still a bit scary to me that one of the lynchpins of Python Data Processing remains, to some extent, to be this page.


Moreover, the builds are delivered over HTTP and the .whl are not signed ...


You only need it if you cannot use a *nix?

So a lynchpins for the Windows Python Data Processing community?


Sadly not everyone can live in a *nix environment. Supporting Windows developers is a pretty damn big lynchpin...


With the new Windows Subsystem for Linux from Windows 10 you can install Ubuntu over Windows from the Windows Store. It is perfect for Data Sciences.


I think Gohlkes should be rewarded for his work. I attribute his efforts to save hours on mine. Thank you for all you've done.


I wish he would let people help mirror them. Instead the links are all obfuscated and you cannot automate downloads which complicates installation.


Yes, the number of times I've run to BLAS/Lapack errors on Windows with Python were countless. I'm glad to see this has been fixed though I moved to Unix OSes a few years back.


could've installed anaconda though


Finally I can uninstall `miniconda` and use the official python+pip.


What’s wrong with anaconda? It’s a good distribution for the science stack.


I've pretty much given up on pip/pip-env on Windows. Anaconda has given me zero problems (even outside of scientific computing) and does everything that I wanted pip/pip-env to do, so I see no reason to switch back.


Even on *nix, anaconda is a godsend for those who don't have sudo and have to install everything in their home directories, without spending a godawful amount of time digging themselves out of dependency hell.


`pip install --user` has been working nicely for a while.


That only works for the python side of things, right?


that's a good point, because conda isn't just python.


Maybe I'm new, but I've not had problems with `venv/bin/pip install` either.


sure. if you have pip.


`python get-pip.py --user`, then... (via https://pip.pypa.io/en/stable/installing/#id4)


if only that had worked.


I used pip & virtualenv on linux constantly. On my new windows laptop, anaconda is preinstalled, so I thought I'd just use that, since I heard from everywhere that it was so good. I ran into issues with it requiring admin privileges or other windows permission issues.

I am no using python 3.6 (which comes with anaconda) create a venv with `python -m venv venv` and I couldn't be happier. Numpy and scipy wheels seem to work (although I needed an `--only-binary scipy` flag added, or something of the like).


Congratulations!

Among many others, AI researchers all over the planet owe a debt of gratitude to these guys. Python dominates in AI research, and deep learning in particular, in part because it has a wealth of libraries like Scipy.

So, THANK YOU!


> Many of us are a bit perfectionist, and therefore are reluctant to call something "1.0" because it may imply that it's "finished" or "we are 100% happy with it".

I always love to see this attitude/approach in an open source project, especially in such a mature one like Scipy. Kudos to the team!


They could adapt TeX versioning scheme (latest release is 3.14159265).


I've also seen a similar scheme on this lib. Each release just keeps adding a nine. They're at version 0.999999999 now:

https://pypi.python.org/pypi/html5lib


That's mostly a horrible accident though.

In retrospect what we should have done is released version 1.0 about 7 years ago when the parser was more or less fully compliant with the spec at the time and just been prepared to bump the version for backwards-incompatible changes (note that at the time the conventions of semver were not as widely adopted as they are now).

What actually happened was we planned a 1.0 release, and so made a series of "final" pre-1.0 version numbers like 0.90, 0.99,etc. and then kept finding one more thing to do, coupled with a decreasing amount of time working on it, and finally spec changes that we didn't keep up with. There certainly wasn't any TeX-like intent to converge on a number, just a series of decisions that seemed reasonable at the time, but had an overall effect that was ridiculous.


So by adding a digit each version, it's technically using a unary numeral system.


Lots of related projects getting that long overdue major version number now. SymPy 1.0 was released last year and I released mpmath 1.0 less than a month ago, both projects 10 years old. SciPy's 16 years is a notch more impressive, though. Congratulations, and thanks to the SciPy developers for their hard work!


Scipy and numpy are both awesome. But I've always wondered what causes so many widely used projects to hold off for so long on the 1.0 release?


Probably this:

Many of us are a bit perfectionist, and therefore are reluctant to call something "1.0" because it may imply that it's "finished" or "we are 100% happy with it". This is normal for many open source projects, however that doesn't make it right. (from the post)


Didn't catch that, thanks.




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: