When a new version of Pyramid is released, it will sometimes deprecate a feature or remove a feature that was deprecated in an older release. When features are removed from Pyramid, applications that depend on those features will begin to break. This chapter explains how to ensure your Pyramid applications keep working when you upgrade the Pyramid version you're using.
The Pyramid core team is conservative when it comes to removing features. We don't remove features unnecessarily, but we're human and we make mistakes which cause some features to be evolutionary dead ends. Though we are willing to support dead-end features for some amount of time, some eventually have to be removed when the cost of supporting them outweighs the benefit of keeping them around, because each feature in Pyramid represents a certain documentation and maintenance burden.
Deprecation and removal policy¶
When a feature is scheduled for removal from Pyramid or any of its official add-ons, the core development team takes these steps:
Using the feature will begin to generate a DeprecationWarning, indicating the version in which the feature became deprecated.
A note is added to the documentation indicating that the feature is deprecated.
A note is added to the Pyramid Change History about the deprecation.
When a deprecated feature is eventually removed:
The feature is removed.
A note is added to the Pyramid Change History about the removal.
Features are never removed in micro releases. They are only removed in minor and major releases. Deprecated features are kept around for at least three minor releases from the time the feature became deprecated. Therefore, if a feature is added in Pyramid 1.0, but it's deprecated in Pyramid 1.1, it will be kept around through all 1.1.X releases, all 1.2.X releases and all 1.3.X releases. It will finally be removed in the first 1.4.X release.
Sometimes features are "docs-deprecated" instead of formally deprecated. This means that the feature will be kept around indefinitely, but it will be removed from the documentation or a note will be added to the documentation telling folks to use some other newer feature. This happens when the cost of keeping an old feature around is very minimal and the support and documentation burden is very low. For example, we might rename a function that is an API without changing the arguments it accepts. In this case, we'll often rename the function, and change the docs to point at the new function name, but leave around a backwards compatibility alias to the old function name so older code doesn't break.
"Docs deprecated" features tend to work "forever", meaning that they won't be removed, and they'll never generate a deprecation warning. However, such changes are noted in the Pyramid Change History, so it's possible to know that you should change older spellings to newer ones to ensure that people reading your code can find the APIs you're using in the Pyramid docs.
Python support policy¶
At the time of a Pyramid version release, each supports all versions of Python through the end of their lifespans. The end-of-life for a given version of Python is when security updates are no longer released.
Python 3.7 Lifespan 2023-06-27.
Python 3.8 Lifespan 2024-10.
Python 3.9 Lifespan 2025-10.
Python 3.10 Lifespan 2026-10.
To determine the Python support for a specific release of Pyramid, view its
tox.ini file at the root of the repository's version.
Consulting the change history¶
Your first line of defense against application failures caused by upgrading to
a newer Pyramid release is always to read the Pyramid Change History to find the
deprecations and removals for each release between the release you're currently
running and the one to which you wish to upgrade. The change history notes
every deprecation within a
Deprecation section and every removal within a
Backwards Incompatibilities section for each release.
The change history often contains instructions for changing your code to avoid deprecation warnings and how to change docs-deprecated spellings to newer ones. You can follow along with each deprecation explanation in the change history, simply doing a grep or other code search to your application, using the change log examples to remediate each potential problem.
Testing your application under a new Pyramid release¶
Once you've upgraded your application to a new Pyramid release and you've remediated as much as possible by using the change history notes, you'll want to run your application's tests (see Run the tests) in such a way that you can see DeprecationWarnings printed to the console when the tests run.
python -Wd setup.py test -q
-Wd argument tells Python to print deprecation warnings to the console.
See the Python -W flag documentation for more
As your tests run, deprecation warnings will be printed to the console explaining the deprecation and providing instructions about how to prevent the deprecation warning from being issued. For example:
python -Wd setup.py test -q # .. elided ... running build_ext /home/chrism/projects/pyramid/env27/myproj/myproj/views.py:3: DeprecationWarning: static: The "pyramid.view.static" class is deprecated as of Pyramid 1.1; use the "pyramid.static.static_view" class instead with the "use_subpath" argument set to True. from pyramid.view import static . ---------------------------------------------------------------------- Ran 1 test in 0.014s OK
In the above case, it's line #3 in the
myproj.views module (
pyramid.view import static) that is causing the problem:
1from pyramid.view import view_config 2 3from pyramid.view import static 4myview = static('static', 'static')
The deprecation warning tells me how to fix it, so I can change the code to do things the newer way:
1from pyramid.view import view_config 2 3from pyramid.static import static_view 4myview = static_view('static', 'static', use_subpath=True)
When I run the tests again, the deprecation warning is no longer printed to my console:
python -Wd setup.py test -q # .. elided ... running build_ext . ---------------------------------------------------------------------- Ran 1 test in 0.014s OK
My application doesn't have any tests or has few tests¶
If your application has no tests, or has only moderate test coverage, running tests won't tell you very much, because the Pyramid codepaths that generate deprecation warnings won't be executed.
In this circumstance, you can start your application interactively under a
server run with the
PYTHONWARNINGS environment variable set to
On Unix, you can do that via:
PYTHONWARNINGS=default $VENV/bin/pserve development.ini
On Windows, you need to issue two commands:
set PYTHONWARNINGS=default Scripts\pserve development.ini
At this point, it's ensured that deprecation warnings will be printed to the console whenever a codepath is hit that generates one. You can then click around in your application interactively to try to generate them, and remediate as explained in Testing your application under a new Pyramid release.
See the PYTHONWARNINGS environment variable documentation or the Python -W flag documentation for more information.
Upgrading to the very latest Pyramid release¶
When you upgrade your application to the most recent Pyramid release, it's advisable to upgrade step-wise through each most recent minor release, beginning with the one that you know your application currently runs under, and ending on the most recent release. For example, if your application is running in production on Pyramid 1.2.1, and the most recent Pyramid 1.3 release is Pyramid 1.3.3, and the most recent Pyramid release is 1.4.4, it's advisable to do this:
Upgrade your environment to the most recent 1.2 release. For example, the most recent 1.2 release might be 1.2.3, so upgrade to it. Then run your application's tests under 1.2.3 as described in Testing your application under a new Pyramid release. Note any deprecation warnings and remediate.
Upgrade to the most recent 1.3 release, 1.3.3. Run your application's tests, note any deprecation warnings, and remediate.
Upgrade to 1.4.4. Run your application's tests, note any deprecation warnings, and remediate.
If you skip testing your application under each minor release (for example if you upgrade directly from 1.2.1 to 1.4.4), you might miss a deprecation warning and waste more time trying to figure out an error caused by a feature removal than it would take to upgrade stepwise through each minor release.