2 Commits

Author	SHA1	Message	Date
AJ	08171767a0	Port to Django 5.2 (#635) ... * Port to Django 5.2 * Pin pluggy to 1.2.0 Any newer and the mystery importlib metadata error appears. Weird! >_> * Update for premailer changed default * Update view logic for is_ajax being changed to a template context processor * Port a few more tests to pytest proper Having two distinct test flavours is giving me a headache * Version 1 dockerfile Makes a VERY big image, I suspect we can optimise this a lot... * Optimise dockerfile a little lot a bit * fix(users): change logout link to POST request * fix(tests): fix some syntax errors in test code still got lots of failed tests :( * fix: replace deprecated Django APIs * ci: update Dockerfile * ci: update Dockerfile * fix(dependencies+tests): update EoL/vulnerable dependencies and improve test stability * Upgrade Python from 3.10 to 3.12 * Update frontend dependencies, replace node-sass with sass * chore(logging): ignore dangling obj reference warning from pypdf * ci: update compose.yml for prod deployment - Add Nginx as a reverse proxy - Add cert-selfsign.sh for generating self-signed certificates - Add certbot-issue.sh for Let's Encrypt certificate issuance and renewal - Add cron-install.sh and cron-uninstall.sh for system cron management - Add .env.example as an environment variable template * style: reformat code * ci: add more sleep trying to pass tests * ci: ignore browser-based tests during ci testing * fix: create home for the new user during Dockerfile building * chore: remove heroku conf file * fix: RIGS not franken anymore * fix: restore is_ajax as a boolean and split out the context processor * test: remove Selenium interaction tests and their CI workarounds * fix: harden production settings (CSRF origins, env casts, ADMINS) and drop dead debug toolbar code * deploy: serve static files from nginx, pin PGDATA, and make cert scripts loud on failure * build: switch .dockerignore to an allowlist * build: drop unused dependencies, soft-pin the rest and target Python 3.14 * build: move image to Python 3.14 / Node 24 and force DEBUG off * ci: add dependabot config for uv, npm, docker and actions * build: narrow Sass deprecation silencing to @import and require Node 24 * build: provide placeholder env for collectstatic now that DEBUG is off in the image * deploy: add plain-HTTP nginx config for local development and use it in compose * deploy: remove self-signed cert script * build: make the image multi-arch with official node and python base images * ci: lint and test on PRs, build the image on PRs and push it to GHCR on master * docs: add local running and sample data instructions * fix: report a form error instead of crashing when big power has no Power MIC * fix: upgrade pypdf and urllib3 to patched releases * fix: apply non-breaking npm audit fixes * fix: strip whitespace in ALLOWED_HOSTS and drop misleading HSTS preload * deploy: add a pyrigs healthcheck and make nginx wait for it * build: pin the uv image version * fix: correct the check-in person picker condition and use BeautifulSoup's string argument * refactor: replace unique_together with UniqueConstraint * build: replace pycodestyle with ruff and fix what it found Removes unused imports and variables, and fixes a few real problems it surfaced: - EventCheckIn.active() referenced an undefined name and raised NameError; it now returns whether the check-in has no end time - RIGS.admin defined EventChecklistAdmin twice; the second is now PowerTestRecordAdmin - RIGS/tests/conftest.py used date/timedelta without importing them - the signal-registering imports in apps.py are kept with noqa pycodestyle config in setup.cfg is dropped. * style: format the codebase with ruff * style: normalise line endings, trailing whitespace and end-of-file newlines * ci: run ruff and file hygiene through prek, and document it --------- Co-authored-by: Hang <me@hangxu.me> Co-authored-by: Joe Banks <joe@jb3.dev>	2026-10-01 23:43:41 +01:00
Arona Jones	2bf0175786	Toolchain/Dependency Upgrade (#418) ... * Upgrade to heroku-20 stack * Move some gulp deps to dev rather than prod * npm upgrade * Fix audit time check in asset audit test * Attempt at parallelising tests where possible * Add basic calendar button test Mainly to pickup on FullCalendar loading errors * Upgrade python deps * Tends to help if I push valid yaml * You valid now? * Fix whoops in requirements.txt * Change python ver * Define service in coveralls task * Run parallelised RIGS tests as one matrix job * Update python version in tests * Cache python dependencies Should majorly speedup parallelillelelised testing * Purge old vagrant config * No Ruby compass bodge, no need for rubocop! * Purge old .idea config * Switch to gh-a artifact uploading instead of imgur 'hack' For test failure screenshots. Happy now @mattysmith22? ;p * Oops, remove unused import * Exclude tests from the coverage stats Seems to be artifically deflating our stats * Refactor asset audit tests with better selectors Also fixed a silly title error with the modal * Add title checking to the slightly insane assets test * Fix unauth test to not just immediately pass out * Upload failure screenshots as individual artifacts not a zip Turns out I can't unzip things from my phone, which is a pain * Should fix asset test on CI * What about this? * What about this? Swear I spend my life jiggerypokerying the damn test suite... * Does this help the coverage be less weird? * Revert "Does this help the coverage be less weird?" This reverts commit 39ab9df836. * Use pytest as our test runner for better parallelism Also rewrote some asset tests to be in the pytest style. May do some more. Some warnings cleaned up in the process. * Bah, codestyle * Oops, remove obsolete if check * Fix screenshot uploading on CI (again) * Try this way of parallel coverage * Add codeclimate maintainability badge * Remove some unused gulp dependencies * Run asset building serverside * Still helps if I commit valid YAML * See below * Different approach to CI dependencies * Exclude node_modules from codestyle * Does this work? * Parallel parallel builds were giving me a headache, try this * Update codeclimate settings, purge some config files * Well the YAML was *syntactically* valid.... * Switch back to old coveralls method * Fix codeclimate config, mark 2 * Attempt to bodge asset test * Oops, again Probably bedtime.. * Might fix heroku building * Attempt #2 at fixing heroku * Belt and braces approach to coverage * Github, you need a Actions YAML validator! * Might fix actions? * Try ignoring some third party deprecation warnings * Another go at making coverage show up * Some template cleanup * Minor python cleanup * Import optimisation * Revert "Minor python cleanup" This reverts commit 6a4620a2e5. * Add format arg to coverage command * Ignore test directories from Heroku slug * Maybe this works to purge deps postbuild * Bunch of test refactoring * Restore signals import, screw you import optimisation * Further template refactoring * Add support for running tests with geckodriver, do this on CI * Screw you codestyle * Disable firefox tests for now That was way more errors than I expected * Run cleanup script from the right location * Plausibly fix tests * Helps if I don't delete the pipeline folder prior to collectstatic * Enable whitenoise * Can I delete pipeline here? * Allow seconds difference in assert_times_equal * Disable codeclimate * Remove not working rm command * Maybe this fixes coverage? * Try different coverage reporter * Fix search_help to need login * Made versioning magic a bit less expansive We have more apps than I thought... * Fix IDI0T error in Assets URLS * Refactor 'no access to unauthed' test to cover all of PyRIGS * Add RAs/Checklists to sample data generator * Fix some HTML errors in templates Which apparently only Django's HTML parser cares about, browsers DGAF... * Port title test to project level * Fix more HTML * Fix cable type detail	2021-01-31 04:05:33 +00:00