Configured the permitted_classes option of YAML::load to include the Date class #13
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Owner
|
Hi, thanks for the PR, am surprised anyone still uses this tbh. I will give it a once over later today.. Also noticed some deps need checking alongside from dependbot.. So will merge and push a updated gem later.. |
kingdonb
pushed a commit
to kingdonb/stats-tracker-ghcr
that referenced
this pull request
Jun 22, 2023
mort666/app_version#13 Signed-off-by: Kingdon Barrett <kingdon@weave.works>
|
I just stumbled on this gem, and it's neat! Does exactly what I wanted, but better than I was looking to do it on my own... I'm using @afaraldo's branch for now and it's working with Ruby 3.1.4 as well 👍 |
kingdonb
pushed a commit
to kingdonb/link-anchor-checker
that referenced
this pull request
Aug 7, 2023
Signed-off-by: Yebyen <yebyen@gmail.com> Highlight the README of lib/ Signed-off-by: Yebyen <yebyen@gmail.com> run 'make test' Signed-off-by: Kingdon Barrett <kingdon@weave.works> Upgrade to Ruby 3.1.4 Signed-off-by: Kingdon Barrett <kingdon@weave.works> rewrite my_wasmer to handle several caches one for every project/repo/package (image == package) Signed-off-by: Kingdon Barrett <kingdon@weave.works> git commit Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup formatting when we call create_leaf Signed-off-by: Kingdon Barrett <kingdon@weave.works> we did it! Signed-off-by: Kingdon Barrett <kingdon@weave.works> delete things we didn't need I did this commit on stage Signed-off-by: Kingdon Barrett <kingdon@weave.works> initial empty commit Signed-off-by: Kingdon Barrett <kingdon@weave.works> scaffold Gemfile + README Signed-off-by: Kingdon Barrett <kingdon@weave.works> .keep files Signed-off-by: Kingdon Barrett <kingdon@weave.works> database config with superuser Signed-off-by: Kingdon Barrett <kingdon@weave.works> app, test, bin, public Signed-off-by: Kingdon Barrett <kingdon@weave.works> delete git_hub_org.rb Signed-off-by: Kingdon Barrett <kingdon@weave.works> add the Kubernetes YAML per the cli docs $ be ruby cli Commands: cli controller ORG # Create a Project for the GitHub ORG and start reconciling it cli help [COMMAND] # Describe available commands or one specific command We implement this line: Create a Project for the GitHub ORG and start reconciling it --- if the Project hasn't been created yet, else just reconcile everything The controller reconciles both Projects and Leaves It also creates Leaf from each project, and each Leaf executes the Wasm as well. The status on the Leaf gets updated then. In the next iteration of this project iterator, we'll make sure that Leaf and Project log their activity back to the database with Rails. Signed-off-by: Kingdon Barrett <kingdon@weave.works> ensure that GithubOrg is created for fluxcd Signed-off-by: Kingdon Barrett <kingdon@weave.works> Alright so the leaves create a db footprint Now we can do after-the-fact measurements of what the result of our Kubernetes operator things have been, without putting further strain on our Kubernetes API service to build out the analytics. Thanks, Rails! Signed-off-by: Kingdon Barrett <kingdon@weave.works> add run fiber to each database model When we create a database row, we can watch our associated Kubernetes resource and do whatever it is we're supposed to do to make sure that it becomes Ready (then update accordingly and mark it as Ready, I think!) The goal is for the top-level fiber GithubOrg to count how many Wasm modules it fires. It should then wait until all its Leaves (Packages) have been created, register more health checks for each of those, and only mark the top-level GithubOrg as ready when all of its Packages are finished. Then, we can take the Measurement snapshot and shut it down! It remains subtly unclear how each of these health checks should work, and how much responsibility each should have. Fibers should enable us to avoid worrying too much about blocking and non-blocking operations, but must keep the order straight else it may yet remain at risk of deadlock. Parents keep count of their children, they care for (wait for) each one to appear inside of the database. When they have finished their useful life, the parent receives the signal to die, which triggers the delete method. All meaningful cleanup happens there, including sending the terminate signal to any children (by deleting their Leaves, which has another delete method to trigger, and each child cleans up after itself.) We may consider the delete method as the place to push metrics, with this pattern in place the path to "Serverless" is almost clear ahead! Signed-off-by: Kingdon Barrett <kingdon@weave.works> increase the test difficulty it has to handle 64 bit integers now (!!) Signed-off-by: Kingdon Barrett <kingdon@weave.works> migrate download_count to bigint Signed-off-by: Kingdon Barrett <kingdon@weave.works> Upgrade download_count field to a bigint Signed-off-by: Kingdon Barrett <kingdon@weave.works> Completes in about 20s fresh packages count: 0 (expecting 16) ####### fresh packages count: 7 (expecting 16) ####### final packages count: 16 (expecting 16) ####### The overall time spent is close to 20s now, but it will be much faster on a Kubernetes control plane that is optimized and a better network. The main delay is most likely due to the single-threadedness of Ruby fibers. The concurrency model is a good experience, even if performance can suffer! Signed-off-by: Kingdon Barrett <kingdon@weave.works> remove instances of 'fluxcd' magic string Signed-off-by: Kingdon Barrett <kingdon@weave.works> add placeholders for conditions This should make it possible for children to record for their loving parents what are the current results of their reconciling attempts. It may make some new requirements on parents, for instance populating the initial status with a reasonable structure with the required items in order to validate the create/apply time resource when it is new (?) No, I think you can just create a new resource without any Conditions and that is "Normal" – it's up to the reconciler to create the Condition items and reflect any changes with Condition updates to show an observed generation for that condition. Signed-off-by: Kingdon Barrett <kingdon@weave.works> add some associations Signed-off-by: Kingdon Barrett <kingdon@weave.works> add basic updating on run I think we don't really have to implement kstatus conditions to get this over the finish line Signed-off-by: Kingdon Barrett <kingdon@weave.works> It has some version of health checking It is neither Kstatus Ready true/false state, nor Conditions We are not consistently using the logger either, because reasons Signed-off-by: Kingdon Barrett <kingdon@weave.works> at the end of the demo both GithubOrg and Packages should have received an update, so their updated_at reflects the last run time. Signed-off-by: Kingdon Barrett <kingdon@weave.works> clean up migration history packages.download_count has always been bigint Signed-off-by: Kingdon Barrett <kingdon@weave.works> separate database connections and separate processes with `foreman start` Signed-off-by: Kingdon Barrett <kingdon@weave.works> reverse order of operations for Sample.ensure run the sample insert script before we start the reconciler (since we're not doing concurrency, the reconciler will never yield back to us, so we need to take care of this ahead of time) Signed-off-by: Kingdon Barrett <kingdon@weave.works> Procfile Signed-off-by: Kingdon Barrett <kingdon@weave.works> no concurrency Signed-off-by: Kingdon Barrett <kingdon@weave.works> make some enhancements to no-concurrency mode Signed-off-by: Kingdon Barrett <kingdon@weave.works> make sure we sleep long enough to delete the leaves all are not getting deleted when we exit this fast Signed-off-by: Kingdon Barrett <kingdon@weave.works> solidify improvements Signed-off-by: Kingdon Barrett <kingdon@weave.works> display count while garbage collecting Signed-off-by: Kingdon Barrett <kingdon@weave.works> additionalPrinterColumns Signed-off-by: Kingdon Barrett <kingdon@weave.works> fine tuning Signed-off-by: Kingdon Barrett <kingdon@weave.works> proj, l (shortnames) Signed-off-by: Kingdon Barrett <kingdon@weave.works> make (time foreman start) Signed-off-by: Kingdon Barrett <kingdon@weave.works> set more consistent use of time Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup db schema Signed-off-by: Kingdon Barrett <kingdon@weave.works> record measurement finally Signed-off-by: Kingdon Barrett <kingdon@weave.works> loosen time constraints Signed-off-by: Kingdon Barrett <kingdon@weave.works> implement conditions into concurrency sad news: it doesn't appear to be any faster (yet) Signed-off-by: Kingdon Barrett <kingdon@weave.works> I think I have understood the YAML::Store cache Signed-off-by: Kingdon Barrett <kingdon@weave.works> add another sleep 5 We haven't got it quite right, if we're exiting the health check before leaves are finished, but it doesn't hurt to sleep after the measurement! Signed-off-by: Kingdon Barrett <kingdon@weave.works> don't emit so many events also, table sharding experiment for now (it failed) Signed-off-by: Kingdon Barrett <kingdon@weave.works> it isn't slower but it isn't faster... but it implements conditions! and it models fast feedback, among other good qualities Signed-off-by: Kingdon Barrett <kingdon@weave.works> let's run "make" in CI and ensure that stat.wasm gets committed into the resulting Docker build Signed-off-by: Kingdon Barrett <kingdon@weave.works> base Signed-off-by: Kingdon Barrett <kingdon@weave.works> add Makefile to dockerignore Signed-off-by: Kingdon Barrett <kingdon@weave.works> update Dockerfile to use pushed base Signed-off-by: Kingdon Barrett <kingdon@weave.works> don't rebuild base every time docker is built Signed-off-by: Kingdon Barrett <kingdon@weave.works> add foreman to base Signed-off-by: Kingdon Barrett <kingdon@weave.works> add 'gems' target for Gemfile cargo build (wasmer) requires the interactive shell for some reason during bundle install Signed-off-by: Kingdon Barrett <kingdon@weave.works> do gem caching Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup Docker build Signed-off-by: Kingdon Barrett <kingdon@weave.works> Docker build fixup Signed-off-by: Kingdon Barrett <kingdon@weave.works> tests passing try break some stuff (hopefully this doesn't break anything) Signed-off-by: Kingdon Barrett <kingdon@weave.works> exclude cache/* from git Signed-off-by: Kingdon Barrett <kingdon@weave.works> add workflow to build a clean cache Signed-off-by: Kingdon Barrett <kingdon@weave.works> set up workflows for building cache + cleaning Signed-off-by: Kingdon Barrett <kingdon@weave.works> add empty option Signed-off-by: Kingdon Barrett <kingdon@weave.works> remember to check out code and include language ruby and rust runtimes/build contexts Signed-off-by: Kingdon Barrett <kingdon@weave.works> latest setup-ruby Signed-off-by: Kingdon Barrett <kingdon@weave.works> packages: write Signed-off-by: Kingdon Barrett <kingdon@weave.works> build base Signed-off-by: Kingdon Barrett <kingdon@weave.works> try to get everything right first published "base" image Signed-off-by: Kingdon Barrett <kingdon@weave.works> add target wasm32-wasi Signed-off-by: Kingdon Barrett <kingdon@weave.works> build binaryen (wasm-opt) Signed-off-by: Kingdon Barrett <kingdon@weave.works> base can be a cache image too in case you have just built base for the first time, you should build the gem-cache tag with CACHE_TAG=base next Then you can build gems, and set gems as the new CACHE_TAG for your deploy build from then on. (I think?) Signed-off-by: Kingdon Barrett <kingdon@weave.works> add wabt Signed-off-by: Kingdon Barrett <kingdon@weave.works> setup-wabt@v1.0.0 Signed-off-by: Kingdon Barrett <kingdon@weave.works> only install wabt for base image We are only compiling stat.wasm in the base image Signed-off-by: Kingdon Barrett <kingdon@weave.works> adopt setup-wabt Signed-off-by: Kingdon Barrett <kingdon@weave.works> 1.0.2 setup-wabt Signed-off-by: Kingdon Barrett <kingdon@weave.works> setup-wabt 1.0.3 Signed-off-by: Kingdon Barrett <kingdon@weave.works> setup-wabt 1.0.4 Signed-off-by: Kingdon Barrett <kingdon@weave.works> try adding wabt (wasm-strip) to the path Signed-off-by: Kingdon Barrett <kingdon@weave.works> try fixing wabt installer to use cache too we don't need all the wabt binaries, just the wasm-strip Signed-off-by: Kingdon Barrett <kingdon@weave.works> mkdir local/bin earlier in workflow Signed-off-by: Kingdon Barrett <kingdon@weave.works> cp -r is not specified Signed-off-by: Kingdon Barrett <kingdon@weave.works> don't try building binaryen in our workflow that's insanity, GHA will only provide us with amd64 runners so that's what type of binary we can expect to use, (and Binaryen provides one) We don't really need to do the wasm stuff in multiple architectures since our output stat.wasm is architecture independent (and will likely be compiled again later by cranelift or something else on the user side) Signed-off-by: Kingdon Barrett <kingdon@weave.works> try to reify cache to include stat.wasm when base build is finished, we can store stat.wasm and bring it back as needed for building targets of other stages Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup syntax error / inconsistency Signed-off-by: Kingdon Barrett <kingdon@weave.works> try to mitigate cache Post job cleanup. Warning: Path Validation Error: Path(s) specified in the action for caching do(es) not exist, hence no cache is being saved. there is no env.HOME Signed-off-by: Kingdon Barrett <kingdon@weave.works> set a new default for publish cache tag parameter Signed-off-by: Kingdon Barrett <kingdon@weave.works> add env.IMAGE_TAG Signed-off-by: Kingdon Barrett <kingdon@weave.works> try restoring these with `docker pull` first if the built tags are fresh enough, they can be used as cache as well Signed-off-by: Kingdon Barrett <kingdon@weave.works> try adding bundle config directly in gem cache I think that's the only place where it's going to be effective Signed-off-by: Kingdon Barrett <kingdon@weave.works> To rebuild base this fix is needed Signed-off-by: Kingdon Barrett <kingdon@weave.works> fix some more discrepancies the `gems` image does not pass `bundle check` without this the `deploy` image can now assume that `gems` will pass bundle check, (and so it does not need to mess with /usr/local/bundle/config anymore) The values of BUNDLE_PATH, GEM_PATH, and GEM_HOME ought to be set by default to /usr/local/bundle, they are kept here for reference in case of nasal demons or whatever. Signed-off-by: Kingdon Barrett <kingdon@weave.works> try caching again I guess we don't build deploy from gem-cache, but from gems Signed-off-by: Kingdon Barrett <kingdon@weave.works> The End of Caching Signed-off-by: Kingdon Barrett <kingdon@weave.works> and a nod to provenance belongs here Signed-off-by: Kingdon Barrett <kingdon@weave.works> some of these instructions are wrong let's fix them... (only fixes some of the wrong things 🤞🤡) Signed-off-by: Kingdon Barrett <kingdon@weave.works> let's try to make local dev work again I know this worked before, even if we can't always depend on CI we should be able to get most of what we need from these caches (and use them to quickly bootstrap the local dev environment) Signed-off-by: Kingdon Barrett <kingdon@weave.works> it's faster to build gems locally on arm64 rather than to wait for github actions to do it with amd64 + qemu Signed-off-by: Kingdon Barrett <kingdon@weave.works> it is faster to cross-build locally it is much faster to build gems-base locally, then tag gem-cache and make docker Pushing layers takes a lot longer than anything else. It ran bundle install and was finished in a second. It did not need Rust The deploy image should be small and fast It is taking over an hour to build gems as multiarch on GitHub, but very thankfully we don't need to build the gems often. We can schedule it as an overnight job that happens once a week, or more like twice a month! Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup README to be true according to what we learn Signed-off-by: Kingdon Barrett <kingdon@weave.works> we have provided for stat.wasm so `make lib` is not needed, only `make test` Also, remember to make a multi-arch Docker `latest` deploy target Signed-off-by: Kingdon Barrett <kingdon@weave.works> trick Dockerfile into thinking stat.wasm is newer Makefile is trying to rebuild stat.wasm, which fails because cargo is no longer present in the deploy context Signed-off-by: Kingdon Barrett <kingdon@weave.works> more explicit forcing Make is noticing that one of the dependencies is missing and using it as an excuse to rebuild stat.wasm (stat.wasm.raw) Signed-off-by: Kingdon Barrett <kingdon@weave.works> make has -t (--touch) for this Signed-off-by: Kingdon Barrett <kingdon@weave.works> add HOST parameter to secrets Signed-off-by: Kingdon Barrett <kingdon@weave.works> use environment variable in ancillary db handlers Signed-off-by: Kingdon Barrett <kingdon@weave.works> add more permission and a new auth mechanism for app/models/measurement singleton Signed-off-by: Kingdon Barrett <kingdon@weave.works> I don't know if we need this ClusterRoleBinding Signed-off-by: Kingdon Barrett <kingdon@weave.works> I guess it was necessary we need at least list CRDs Signed-off-by: Kingdon Barrett <kingdon@weave.works> maybe we didn't Signed-off-by: Kingdon Barrett <kingdon@weave.works> Revert "I guess it was necessary" This reverts commit 1e116b52d64a50e129201110d28877db48ea3257. I guess it was necessary we need at least list CRDs Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup kubernetes default endpoint Signed-off-by: Kingdon Barrett <kingdon@weave.works> add concurrencyPolicy Signed-off-by: Kingdon Barrett <kingdon@weave.works> tighten up clusterrole quite a bit from cluster-admin to list and get CRDs, (which we need in order for the client to reflectively discover leaf and project APIs, yay kubeclient!) also, add "update" since it seems to be used by kubeclient as well Signed-off-by: Kingdon Barrett <kingdon@weave.works> in lieu of production deployment for now we'll have this one environment, which gets two hostnames (and the DNS is configured by hand, at DigitalOcean, to match whatever test cluster is in vogue right now...) Signed-off-by: Kingdon Barrett <kingdon@weave.works> I think it's list outside namespace The operator isn't built to be operating confined to a single ns Signed-off-by: Kingdon Barrett <kingdon@weave.works> add events to rbac Signed-off-by: Kingdon Barrett <kingdon@weave.works> add CI build workflow for main branch Signed-off-by: Kingdon Barrett <kingdon@weave.works> add missing "break" statement we should not issue the "extra" delete twice Signed-off-by: Kingdon Barrett <kingdon@weave.works> rescue Date::Error this can be false Signed-off-by: Kingdon Barrett <kingdon@weave.works> that wasn't exactly exiting cleanly it's still supposed to clean up after project gets deleted Signed-off-by: Kingdon Barrett <kingdon@weave.works> decrease the amount of waiting Signed-off-by: Kingdon Barrett <kingdon@weave.works> add hosts to environment development Signed-off-by: Kingdon Barrett <kingdon@weave.works> try to reconcile all leaves on startup Signed-off-by: Kingdon Barrett <kingdon@weave.works> try to introduce debugging, Stalled condition when we detect the resources should have progressed by now, we can set the Stalled condition that will trigger them to be reconciled again While introducing the Stalled condition, I believe I have accidentally solved recovery as well (🥂🚀🎉✅😁) Signed-off-by: Kingdon Barrett <kingdon@weave.works> upgrade Ruby, install rspec Signed-off-by: Kingdon Barrett <kingdon@weave.works> delete vendor/ we won't be vendoring any gems or javascript Signed-off-by: Kingdon Barrett <kingdon@weave.works> bundle update Signed-off-by: Kingdon Barrett <kingdon@weave.works> move things around so upsert method is not so big this shouldn't have changed any of the behavior, but it makes clearer how convoluted our upsert method has already gotten... Signed-off-by: Kingdon Barrett <kingdon@weave.works> factory_bot_rails Signed-off-by: Kingdon Barrett <kingdon@weave.works> implement AR::BaseConnection into leaf_reconciler implement it here first, reuse it everywhere later Signed-off-by: Kingdon Barrett <kingdon@weave.works> factory_bot (bundle install) Signed-off-by: Kingdon Barrett <kingdon@weave.works> bundle update (puma) Signed-off-by: Kingdon Barrett <kingdon@weave.works> attempt to boil the ocean Signed-off-by: Kingdon Barrett <kingdon@weave.works> debugging Signed-off-by: Kingdon Barrett <kingdon@weave.works> reconcile_leaf.rb It works but it crashes @end_line=148, @end_column=47>]:Array (NoMethodError) leafer.1 | leafer.1 | @selector.select(@timeouts.interval) leafer.1 | ^^^^^^^ leafer.1 | from /Users/kingdonb/.rvm/gems/ruby-3.2.2/bundler/gems/fiber_scheduler-442188f8c752/lib/fiber_scheduler.rb:122:in `run_once' leafer.1 | from /Users/kingdonb/.rvm/gems/ruby-3.2.2/bundler/gems/fiber_scheduler-442188f8c752/lib/fiber_scheduler.rb:116:in `run' Signed-off-by: Kingdon Barrett <kingdon@weave.works> exclude leafer from foreman so can run it by hand it crashes at the end, so something still is not right, however it seems like it performs, and we may be able to put this one back in the rotation without any further changes (except this will need to be reverted) Signed-off-by: Kingdon Barrett <kingdon@weave.works> process pool didn't work not sure why, but this version is based on Ruby 3.1.4, it looks like it works damn near every time, it doesn't crash until the end, and it's in perfect shape to be shipped as the cronjob right now Signed-off-by: Kingdon Barrett <kingdon@weave.works> adopt BaseConnection class one last place Signed-off-by: Kingdon Barrett <kingdon@weave.works> revert to rubygems.org kubernetes-operator GitHub will not permit me to pull in the bundle gem from GH Package Registry without authenticating myself, which I cannot easily do inside a Dockerfile, (and we don't need this since we didn't ultimately take the Ruby 3.2 upgrade path for now) Signed-off-by: Kingdon Barrett <kingdon@weave.works> rake app:version $ rake app:version Application version: 0.0.1 (187) by Kingdon Barrett on 2023-06-22 Signed-off-by: Kingdon Barrett <kingdon@weave.works> app_version gem is an oldie but a goodie mort666/app_version#13 Signed-off-by: Kingdon Barrett <kingdon@weave.works> reorganize manifests into bases podinfo is the inspiration for this design, we'll have deploy/bases and deploy/overlays for now (with each environment as a folder in overlays) Signed-off-by: Kingdon Barrett <kingdon@weave.works> add kustomization.yaml for each base Signed-off-by: Kingdon Barrett <kingdon@weave.works> rotate credentials.yml for production Signed-off-by: Kingdon Barrett <kingdon@weave.works> hotfix Signed-off-by: Kingdon Barrett <kingdon@weave.works> rake app:render Signed-off-by: Kingdon Barrett <kingdon@weave.works> lograge (Gemfile) Signed-off-by: Kingdon Barrett <kingdon@weave.works> initializer (lograge) Signed-off-by: Kingdon Barrett <kingdon@weave.works> Health check route (200) Signed-off-by: Kingdon Barrett <kingdon@weave.works> RAILS_ENV production in cronjob Signed-off-by: Kingdon Barrett <kingdon@weave.works> add dev and production overlays change RAILS_ENV variable and deployment namespace Signed-off-by: Kingdon Barrett <kingdon@weave.works> set /healthz as health endpoint Signed-off-by: Kingdon Barrett <kingdon@weave.works> Let's attempt to publish 0.0.1 Signed-off-by: Kingdon Barrett <kingdon@weave.works> checkout before prepare step Signed-off-by: Kingdon Barrett <kingdon@weave.works> set up ruby before rake Signed-off-by: Kingdon Barrett <kingdon@weave.works> must bundle install before rake will work Signed-off-by: Kingdon Barrett <kingdon@weave.works> populate bundler-cache from main branch Signed-off-by: Kingdon Barrett <kingdon@weave.works> bundle lock --add-platform x86_64-linux Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release v0.0.1 Signed-off-by: Kingdon Barrett <kingdon@weave.works> hotfix workflow Signed-off-by: Kingdon Barrett <kingdon@weave.works> hotfix workflow #2 Signed-off-by: Kingdon Barrett <kingdon@weave.works> update production overlay (0.0.1) Signed-off-by: Kingdon Barrett <kingdon@weave.works> we can fetch GithubOrg before health check If we fetch GithubOrg too early, it won't tell us how many packages to expect here Signed-off-by: Kingdon Barrett <kingdon@weave.works> bump version in template, overlay, rake app:render Signed-off-by: Kingdon Barrett <kingdon@weave.works> update checks in README (#29) Signed-off-by: Kingdon Barrett <kingdon@weave.works> add Execute workflow (#30) Signed-off-by: Kingdon Barrett <kingdon@weave.works> kubectl apply -k Signed-off-by: Kingdon Barrett <kingdon@weave.works> Add a "publish wasm" step to tag release workflow Signed-off-by: Kingdon Barrett <kingdon@weave.works> bump version in template, overlay, rake app:render Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup publish workflow Signed-off-by: Kingdon Barrett <kingdon@weave.works> change execute workflow to fetch stat.wasm Signed-off-by: Kingdon Barrett <kingdon@weave.works> set release_name Docs here suggest that release_name is the tag by default: https://github.com/svenstaro/upload-release-action#input-variables IME it looks like the commit message of the tagged commit is actually being used, which can make some ugly release messages (at least until the release commits are being generated by a release manager robot!) Signed-off-by: Kingdon Barrett <kingdon@weave.works> add sbom and provenance, add cosign Signed-off-by: Kingdon Barrett <kingdon@weave.works> add release machinery to Makefile Signed-off-by: Kingdon Barrett <kingdon@weave.works> echo current version during 'make prerel' Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.0 make prerel Signed-off-by: Kingdon Barrett <kingdon@weave.works> fix minor issue in publish-tag workflow Now it should not print "vrefs/tags/0.1.0" in the release notes title Signed-off-by: Kingdon Barrett <kingdon@weave.works> Do these docker pull commands have any effect I don't think so Signed-off-by: Kingdon Barrett <kingdon@weave.works> I hope these docker pull commands had no effect Signed-off-by: Kingdon Barrett <kingdon@weave.works> signatures slow things down pretty badly if we want our development builds to return in less than 5 minutes, they'll have to be unsigned and without SBOMs or provenance attestation (I am not sure I had this set up correctly to begin with, there are many warnings in the output from "syft", I don't know how to read it at all.) Signed-off-by: Kingdon Barrett <kingdon@weave.works> add builder id to publish tasks Signed-off-by: Kingdon Barrett <kingdon@weave.works> Try setting up cache manually The Rust setup action considers the package root (.) to be the workspace, in our case it's actually lib/stat It is explained here in the docs that another action is used for the cache config, and it's possible to disable the auto-configured thusly and replace it with your own incantations as this commit does. Hope this works 🤞🚀 Signed-off-by: Kingdon Barrett <kingdon@weave.works> We'll only use the rust cache in the base image We're not building that wasm except in the base image, and at release time Signed-off-by: Kingdon Barrett <kingdon@weave.works> note: base still not publishing SBOM or signatures gems, gem-cache, and base image are all unversioned and unsigned Signed-off-by: Kingdon Barrett <kingdon@weave.works> don't need either of these Don't need Ruby unless we're building something outside of the Docker container, and... We don't need the id-token permission if we're not signing anything Signed-off-by: Kingdon Barrett <kingdon@weave.works> mirror caching configuration from publish in the publish-tag workflow (this should make building rust modules in the "publish-tag" workflow much faster!) Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.1 Signed-off-by: Kingdon Barrett <kingdon@weave.works> de-emphasize the "latest" image nobody should be pulling `ghcr.io/kingdonb/stats-tracker-ghcr:latest` Now that we are doing versioned releases, set imagePullPolicy to IfNotPresent and use the versioned artifacts, (or use the canary release if you're absolutely in a dev environment!) Signed-off-by: Kingdon Barrett <kingdon@weave.works> change from latest to canary in manifests Signed-off-by: Kingdon Barrett <kingdon@weave.works> set imagePullPolicy in production we should override this in the dev environment, but I want this change to take effect in production so I'll write it there for greater clarity Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.2 Signed-off-by: Kingdon Barrett <kingdon@weave.works> let's change the definition of prod Signed-off-by: Kingdon Barrett <kingdon@weave.works> add two new CRDs for new reconcilers we will now reconcile PackageVersions when we want to know about the versions of a package, and PackageVersions::VersionLeaf is similar to Project::Leaf in the existing controller pair. Let's also finish implementation of Conditions here (in the following commits) Signed-off-by: Kingdon Barrett <kingdon@weave.works> these are the new entrypoints we'll need Signed-off-by: Kingdon Barrett <kingdon@weave.works> after reviewing the site structure, I think... we can probably do this without a VersionLeaf Signed-off-by: Kingdon Barrett <kingdon@weave.works> move patches related to ImagePullPolicy into dev the default ImagePullPolicy is IfNotPresent Signed-off-by: Kingdon Barrett <kingdon@weave.works> let's try to do this without any others we only need one reconciler because there is only one layer deep (there's no need to fetch multiple pages for each package, and we already know which packages we want to monitor versions for, so there's no need to rewrite GithubOrg or "Project" because there is no second fetch.) This also means we're not writing a second Wasm, lol Signed-off-by: Kingdon Barrett <kingdon@weave.works> ensure sample these are the seed records (this is not a real controller either) Signed-off-by: Kingdon Barrett <kingdon@weave.works> we also will not need this Signed-off-by: Kingdon Barrett <kingdon@weave.works> rails g model Version, VersionMeasurement rails generate model Version package:references version:string download_count:integer rails generate model VersionMeasurement package:references version:references count:integer measured_at:datetime Signed-off-by: Kingdon Barrett <kingdon@weave.works> rails db:migrate Signed-off-by: Kingdon Barrett <kingdon@weave.works> add models Signed-off-by: Kingdon Barrett <kingdon@weave.works> add default tests (rspec, factorybot) Signed-off-by: Kingdon Barrett <kingdon@weave.works> fixup PackageVersion sampler Signed-off-by: Kingdon Barrett <kingdon@weave.works> let's test with multiple PackageVersions Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.3 Signed-off-by: Kingdon Barrett <kingdon@weave.works> some errors in release 0.1.3 Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.4 Signed-off-by: Kingdon Barrett <kingdon@weave.works> last ditch fix effort Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.5 Signed-off-by: Kingdon Barrett <kingdon@weave.works> forgot to do a thing don't skip steps when you're releasing to production, always test in dev Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.6 Signed-off-by: Kingdon Barrett <kingdon@weave.works> measure many versions Signed-off-by: Kingdon Barrett <kingdon@weave.works> don't abort so harshly on alternative sample kinds Signed-off-by: Kingdon Barrett <kingdon@weave.works> Release 0.1.7 Signed-off-by: Kingdon Barrett <kingdon@weave.works>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Now the gem is compatible with ruby 3.2.2
There was a change in psych from 3.1 to 3.2.2 that throws the following error
Tried to load unspecified class: Date (Psych::DisallowedClass)