Mac doesn’t have a preinstalled package manager, but there are a couple of popular package managers you can install. For Python 3.5 with Macports, execute this command in a terminal: sudo port install py35 - numpy py35 - scipy py35 - matplotlib py35 - ipython + notebook py35 - pandas py35 - sympy py35 - nose. The official home of the Python Programming Language. While Javascript is not essential for this website, your interaction with the content will be limited. I aim to install scipy and numpy on Python 2.7. My computer is a Mac version OSX 10.8.5. It is the first time I try to install a package on Python.
As a general-purpose programming language, Python is designed to beused in many ways. You can build web sites or industrial robots or agame for your friends to play, and much more, all using the samecore technology.
Python’s flexibility is why the first step in every Python projectmust be to think about the project’s audience and the correspondingenvironment where the project will run. It might seem strange to thinkabout packaging before writing code, but this process does wonders foravoiding future headaches.
This overview provides a general-purpose decision tree for reasoningabout Python’s plethora of packaging options. Read on to choose the besttechnology for your next project.
Contents
Packages exist to be installed (or deployed), so before you packageanything, you’ll want to have some answers to the deployment questionsbelow:
Who are your software’s users? Will your software be installed byother developers doing software development, operations people in adatacenter, or a less software-savvy group?
Is your software intended to run on servers, desktops, mobileclients (phones, tablets, etc.), or embedded in dedicated devices?
Is your software installed individually, or in large deployment batches?
Packaging is all about target environment and deploymentexperience. There are many answers to the questions above and eachcombination of circumstances has its own solutions. With thisinformation, the following overview will guide you to the packagingtechnologies best suited to your project.
You may have heard about PyPI, setup.py, and wheelfiles. These are just a few of the tools Python’s ecosystem providesfor distributing Python code to developers, which you can read about inPackaging and distributing projects.
The following approaches to packaging are meant for libraries andtools used by technical audience in a development setting. If you’relooking for ways to package Python for a non-technical audience and/ora production setting, skip ahead to Packaging Python applications.
A Python file, provided it only relies on the standard library, can beredistributed and reused. You will also need to ensure it’s writtenfor the right version of Python, and only relies on the standardlibrary.
This is great for sharing simple scripts and snippets between peoplewho both have compatible Python versions (such as via email,StackOverflow, or GitHub gists). There are even some entire Pythonlibraries that offer this as an option, such as bottle.py and boltons.
However, this pattern won’t scale for projects that consist ofmultiple files, need additional libraries, or need a specific versionof Python, hence the options below.
If your code consists of multiple Python files, it’s usually organizedinto a directory structure. Any directory containing Python files cancomprise an Import Package.
Because packages consist of multiple files, they are harder todistribute. Most protocols support transferring only one file at atime (when was the last time you clicked a link and it downloadedmultiple files?). It’s easier to get incomplete transfers, and harderto guarantee code integrity at the destination.
So long as your code contains nothing but pure Python code, and youknow your deployment environment supports your version of Python, thenyou can use Python’s native packaging tools to create a sourceDistribution Package, or sdist for short.
Python’s sdists are compressed archives (.tar.gz files)containing one or more packages or modules. If your code ispure-Python, and you only depend on other Python packages, you can gohere to learn more.
If you rely on any non-Python code, or non-Python packages (such aslibxml2 in the case oflxml, or BLAS libraries in thecase of numpy), you will need touse the format detailed in the next section, which also has manyadvantages for pure-Python libraries.
Note
Python and PyPI support multiple distributions providingdifferent implementations of the same package. For instance theunmaintained-but-seminal PIL distribution provides the PIL package, and sodoes Pillow, anactively-maintained fork of PIL!
This Python packaging superpower makes it possible for Pillow to bea drop-in replacement for PIL, just by changing your project’sinstall_requires or requirements.txt.
So much of Python’s practical power comes from its ability tointegrate with the software ecosystem, in particular libraries writtenin C, C++, Fortran, Rust, and other languages.
Not all developers have the right tools or experiences to build thesecomponents written in these compiled languages, so Python created theWheel, a package format designed to ship libraries withcompiled artifacts. In fact, Python’s package installer, pip,always prefers wheels because installation is always faster, so evenpure-Python packages work better with wheels.
Binary distributions are best when they come with source distributionsto match. Even if you don’t upload wheels of your code for everyoperating system, by uploading the sdist, you’re enabling users ofother platforms to still build it for themselves. Default topublishing both sdist and wheel archives together, unless you’recreating artifacts for a very specific use case where you know therecipient only needs one or the other.
Python and PyPI make it easy to upload both wheels and sdiststogether. Just follow the Packaging Python Projectstutorial.
Python’s recommended built-in library and tool packagingtechnologies. Excerpted from The Packaging Gradient (2017).¶
So far we’ve only discussed Python’s native distribution tools. Basedon our introduction, you would be correct to infer these built-inapproaches only target environments which have Python, and anaudience who knows how to install Python packages.
With the variety of operating systems, configurations, and people outthere, this assumption is only safe when targeting a developeraudience.
Python’s native packaging is mostly built for distributing reusablecode, called libraries, between developers. You can piggybacktools, or basic applications for developers, on top of Python’slibrary packaging, using technologies like setuptools entry_points.
Libraries are building blocks, not complete applications. Fordistributing applications, there’s a whole new world of technologiesout there.
The next few sections organize these application packaging optionsaccording to their dependencies on the target environment,so you can choose the right one for your project.
Some types of Python applications, like web site backends and othernetwork services, are common enough that they have frameworks toenable their development and packaging. Other types of applications,like dynamic web frontends and mobile clients, are complex enough totarget that a framework becomes more than a convenience.
In all these cases, it makes sense to work backwards, from theframework’s packaging and deployment story. Some frameworks include adeployment system which wraps the technologies outlined in the rest ofthe guide. In these cases, you’ll want to defer to your framework’spackaging guide for the easiest and most reliable production experience.
If you ever wonder how these platforms and frameworks work under thehood, you can always read the sections beyond.
If you’re developing for a “Platform-as-a-Service” or “PaaS” likeHeroku or Google App Engine, you are going to want to follow theirrespective packaging guides.
“Serverless” frameworks like Zappa
In all these setups, the platform takes care of packaging anddeployment, as long as you follow their patterns. Most software doesnot fit one of these templates, hence the existence of all the otheroptions below.
If you’re developing software that will be deployed to machines youown, users’ personal computers, or any other arrangement, read on.
Python’s steady advances are leading it into new spaces. These daysyou can write a mobile app or web application frontend inPython. While the language may be familiar, the packaging anddeployment practices are brand new.
If you’re planning on releasing to these new frontiers, you’ll want tocheck out the following frameworks, and refer to their packagingguides:
If you are not interested in using a framework or platform, or justwonder about some of the technologies and techniques utilized by theframeworks above, continue reading below.
Pick an arbitrary computer, and depending on the context, there’s a verygood chance Python is already installed. Included by default in mostLinux and Mac operating systems for many years now, you can reasonablydepend on Python preexisting in your data centers or on the personalmachines of developers and data scientists.
Best Python Packages For Machine Learning
Technologies which support this model:
PEX (Python EXecutable)
zipapp (does not help manage dependencies, requires Python 3.5+)
shiv (requires Python 3)
Note
Of all the approaches here, depending on a pre-installedPython relies the most on the target environment. Of course,this also makes for the smallest package, as small assingle-digit megabytes, or even kilobytes.
In general, decreasing the dependency on the target systemincreases the size of our package, so the solutions hereare roughly arranged by increasing size of output.
For a long time many operating systems, including Mac and Windows,lacked built-in package management. Only recently did these OSes gainso-called “app stores”, but even those focus on consumer applicationsand offer little for developers.
Python Software For Mac
Developers long sought remedies, and in this struggle, emerged withtheir own package management solutions, such as Homebrew. The most relevant alternative for Pythondevelopers is a package ecosystem called Anaconda. Anacondais built around Python and is increasingly common in academic,analytical, and other data-oriented environments, even making its wayinto server-oriented environments.
Instructions on building and publishing for the Anaconda ecosystem:
A similar model involves installing an alternative Pythondistribution, but does not support arbitrary operating system-levelpackages:
Computing as we know it is defined by the ability to executeprograms. Every operating system natively supports one or more formatsof program they can natively execute.
There are many techniques and technologies which turn your Pythonprogram into one of these formats, most of which involve embedding thePython interpreter and any other dependencies into a single executablefile.
This approach, called freezing, offers wide compatiblity andseamless user experience, though often requires multiple technologies,and a good amount of effort.
A selection of Python freezers:
pyInstaller - Cross-platform
cx_Freeze - Cross-platform
constructor - For command-line installers
py2exe - Windows only
py2app - Mac only
bbFreeze - Windows, Linux, Python 2 only
osnap - Windows and Mac
pynsist - Windows only
Most of the above imply single-user deployments. For multi-componentserver applications, see Chef Omnibus.
An increasing number of operating systems – including Linux, Mac OS,and Windows – can be set up to run applications packaged aslightweight images, using a relatively modern arrangement oftenreferred to as operating-system-level virtualization,or containerization.
These techniques are mostly Python agnostic, because they packagewhole OS filesystems, not just Python or Python packages.
Adoption is most extensive among Linux servers, where the technologyoriginated and where the technologies below work best:
Most operating systems support some form of classical virtualization,running applications packaged as images containing a full operatingsystem of their own. Running these virtual machines, or VMs, is amature approach, widespread in data center environments.
These techniques are mostly reserved for larger scale deployments indata centers, though certain complex applications can benefit fromthis packaging. Technologies are Python agnostic, and include:
VHD, AMI, and other formats
OpenStack - A cloud management system in Python, with extensive VM support
The most all-encompassing way to ship your software would be to shipit already-installed on some hardware. This way, your software’s userwould require only electricity.
Whereas the virtual machines described above are primarily reservedfor the tech-savvy, you can find hardware appliances being used byeveryone from the most advanced data centers to the youngest children.
Embed your code on an Adafruit,MicroPython, or more-powerful hardwarerunning Python, then ship it to the datacenter or your users’homes. They plug and play, and you can call it a day.
The simplified gamut of technologies used to package Python applications.¶
The sections above can only summarize so much, and you might bewondering about some of the more conspicuous gaps.
As mentioned in Depending on a separate software distribution ecosystem above, some operatingsystems have package managers of their own. If you’re very sure of theoperating system you’re targeting, you can depend directly on a formatlike deb (forDebian, Ubuntu, etc.) or RPM (for Red Hat,Fedora, etc.), and use that built-in package manager to take care ofinstallation, and even deployment. You can even use FPM togenerate both deb and RPMs from the same source.
In most deployment pipelines, the OS package manager is just one pieceof the puzzle.
Python Extension Packages For Mac
Virtualenvs havebeen an indispensible tool for multiple generations of Pythondeveloper, but are slowly fading from view, as they are being wrappedby higher-level tools. With packaging in particular, virtualenvs areused as a primitive in the dh-virtualenv tool andosnap, both of which wrapvirtualenvs in a self-contained way.
Python Packages For Machine Learning
For production deployments, do not rely on running pipinstallfrom the Internet into a virtualenv, as one might do in a developmentenvironment. The overview above is full of much better solutions.
Download Python Packages For Mac
The further down the gradient you come, the harder it gets to updatecomponents of your package. Everything is more tightly bound together.
For example, if a kernel security issue emerges, and you’re deployingcontainers, the host system’s kernel can be updated without requiringa new build on behalf of the application. If you deploy VM images,you’ll need a new build. Whether or not this dynamic makes one optionmore secure is still a bit of an old debate, going back to thestill-unsettled matter of static versus dynamic linking.
Python 3 Packages For Machine Learning
Packaging in Python has a bit of a reputation for being a bumpyride. This impression is mostly a byproduct of Python’sversatility. Once you understand the natural boundaries between eachpackaging solution, you begin to realize that the varied landscape isa small price Python programmers pay for using one of the mostbalanced, flexible language available.