All new code requires tests to ensure against regressions.
However, if your PR contains zero code changes, feel free to select the checkmark below to indicate so.
[X] Have you ran tests against this code?
[ ] This PR contains zero code changes.
Description of the Change
Originally this PR aimed to implement nock to achieve true 100% integration tests. By simple mocking the external endpoints we had to hit, setting up a database, and hitting our own backend to see what it would do.
A valiant goal in and of itself, but in doing so, I was able to go ahead and resolve one known bug, another unknown bug, as well as improve the user experience considerably.
[Known Bug] Via a new constructNewPackagePublishData builder (much like the builders we already have) I was able to retire the PackageObject builder as it was clunky, and was the cause of bugs itself. Such that it would incorrectly create versioned data during package publication. Resulting in the incorrect tag tarball being selected for the latest tag tarball. This would mean that while the version number was correct, the tarball would point to the wrong location, causing bad installations.
Fixes #236
[Unknown] On every package publish the backend would try to send out a webhook to Discord. But if the publish failed, then the webhook would crash, causing the server to crash as a whole. This has apparently been happening since we implemented webhooks way back when.
[Improved Experience] The new package data builder does actually some minimal work to validate the data it's going to publish, there still is more that can be done here, but as of now it will validate version information, meaning that if it can't find the correct version info it will return a detailed error to the user, which previously it'd report just "Server Error". Additionally, this new builder has many more safeguards in place to discover required info, such as the owner key which will search the package.jsonname field then name of the provided URL for publication, then the parsed URL of the repository key (supporting both string and object form) and finally grabbing the author field. With this being easily expandable in the future for other values.
Requirements
Filling out the template is required.
All new code requires tests to ensure against regressions.
[X] Have you ran tests against this code?
[ ] This PR contains zero code changes.
Description of the Change
Originally this PR aimed to implement
nockto achieve true 100% integration tests. By simple mocking the external endpoints we had to hit, setting up a database, and hitting our own backend to see what it would do.A valiant goal in and of itself, but in doing so, I was able to go ahead and resolve one known bug, another unknown bug, as well as improve the user experience considerably.
constructNewPackagePublishDatabuilder (much like the builders we already have) I was able to retire thePackageObjectbuilder as it was clunky, and was the cause of bugs itself. Such that it would incorrectly create versioned data during package publication. Resulting in the incorrect tag tarball being selected for the latest tag tarball. This would mean that while the version number was correct, the tarball would point to the wrong location, causing bad installations.Fixes #236
ownerkey which will search thepackage.jsonnamefield then name of the provided URL for publication, then the parsed URL of therepositorykey (supporting both string and object form) and finally grabbing theauthorfield. With this being easily expandable in the future for other values.