meta(changelog): Add changelog for v8.0.0-beta.0 #11606
Merged
Conversation
This file contains 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
It may happen that we have different modules, different versions etc. so the weakmaps may not be in sync. Just storing a non-enumerable property should be the safer choice here. Partially required for next.js 13.4.19 to work....
This can be used to lazy load a pluggable integration.
Usage:
```js
async function getOrLazyLoadFeedback() {
const existing = Sentry.getFeedback(); // check if it has already been installed
if (existing) {
return existing;
}
try {
const feedbackIntegration = await Sentry.lazyLoadIntegration('feedbackIntegration');
client.addIntegration(feedbackIntegration());
} catch(error) {
// this can error, we need to handle this!
}
}
```
Closes #10905
…mentations (#11434) Concrete changes: 1. `addRequestDataToEvent` no longer applies its transaction name to _non-transaction_ events 2. Http instrumentation sets the isolation scope's `transactionName` in the Otel `requestHook` to the unparameterized request URL path (w/o query params or fragments). 3. Express instrumentation sets the isolation scope's `transactionName` in the Otel `spanName` hook to a parameterized route, once a request handler span (e.g. `app.get(...)`) is created. 4. As a consequence of the above, non-transaction events now get assigned `event.transaction` correctly from the scopes.
To ensure we can use this in an isomorphic way. This means we do not rely on this being a mutable object anymore, but we need to ensure to update it (in http integration) on both the span _and_ for the current context.
Next week I'm going to play around with optimizing feedback/replay packages as per what we saw in #11452. In prep for that work I re-organized size limit and made the limits a little more sensible. The size bot comment will be a little messed up because we renamed the fields, but after we merge should be a lot better.
All of the ANR tests were verified over stdout. This PR ensures some of the basic tests actually send events over http to ensure the transport is fully tested.
Also adds tests for this for core, vercel-edge and opentelemetry. NOTE: In core, this does not do anything, as we cannot actually update the isolation scope in browser. Co-authored-by: Luca Forstner <[email protected]>
[Gitflow] Merge master into develop
This updates handling of next.js instrumentation to re-use an isolation scope from a root span. This should ensure we have consistent isolation scopes, no matter if next.js auto creates spans or not.
I want to remove the karma/mocha based tests in the browser package. To accomplish this, I'll be porting 1 test suite a day from the old integration tests to playwright. Today is Day 3: `packages/browser/test/integration/suites/config.js` I was surprised we never had `ignoreErrors` or `denyUrls` tests in playwright, so it's good to get the confidence that everything works here. ref #11084 day 2: #11436
Font sizes were set instead of using config option
…the exported surface area (#11355) You can read this commit-by-commit to watch the refactor unfold The situation is: after #11342 is merged, all the code inside each of the `packages/feedback/src/[core|modal|screenshot]` will be split up into separate integrations. These integrations can't share the same `src/types/*.ts` definitions anymore. Therefore type definitions will need to live in @sentry/types instead. This PR moves all the types, and since they'll be public now in that package, i refactored things to reduce the public surface area and improve names where possible. The only non-type changes in here are where we move `createDialog.ts` and `createInput.ts` into their respective `integration.ts` files, no code changes at all. Related to #11435
These are deps we may want to keep more up to date than we usually do, so generating update PRs for them may be helpful.
When interacting with the Breadcrumd API, I had some trouble finding the correct documentation. I retrieved my changes for the previous [PR](#11119) to make sure the JSDoc is updated with the actual doc.
…ng spans (#11503) Our Http instrumentation didn't clone the isolation scope correctly when the request span is not recording. We early returned if the span was not a server kind span. However, non-recording spans apparently have no span kind, so we wrongfully early returned in the check for non-recording spans.
Next.js provides their own OTel http integration, which conflicts with ours ref #11016 added commit from this PR: #11319 --------- Co-authored-by: Luca Forstner <[email protected]> Co-authored-by: Francesco Novy <[email protected]>
Fixes the Canary tests by ensuring we cache node modules and files in dev-packages as well. In the debugging process, I also moved a few things around - e.g. the playwright dependency stuff now happens in CI, not in the build commands everywhere.
Instead of exporting `_INTERNAL` here, it's OK IMHO to export this. The API is stable enough.
…de request (#11511) Update the scope's `transactionName` in the `sentryHandle` instrumentation that's always invoked when a request is made to the SvelteKit server.
…11413) Modifies `makeOfflineTransport` to allow replay envelopes to be queued when offline and to ensure they are always send in order. To do this we: - Modify the store interface so it has `push`, `unshift` and `shift` methods (BREAKING) - Always queue replay envelopes in the store so there is never more than one in-flight - Always `unshift` failed envelopes so they're always sent in the same order from the queue
…11447) This PR updates the isolation scope's `transactionName` by adding another hook callback in our `setupFastifyErrorHandler` fastify plugin. This is better than accessing the otel span for two reasons: 1. We can always update the transactionName, not just when a span is sampled and recording 2. The `onRequest` hook is executed earlier in the lifecycle than the `preHandler` hook that Otel uses to trigger the `requestHook` callback.
As per #4784 (comment) make sure fastify types use any is that we don't collide with what fastify types expect. Also convert e2e tests to use typescript to validate build.
Instead, users can use `registerSpanErrorInstrumentation` which is now does much less things than before. I opted to not fully remove this yet because I guess the chance that somebody may have been using this is somewhat high, and we can easily continue to have it with no bundle size hit for those who don't need it. We also couldn't really deprecate this in v7 so I think it's OK to leave this in for now.
…nitely (#11578) When experimenting for #11525 I found that `Runtime.evaluate` only returns when the event loop becomes unblocked. This means that we are not sending ANR events if the event loop is blocked indefinitely. This PR adds a timeout that sends the ANR event if we have not been able to evaluate the scope within 5 seconds.
This also forward-ports the tests from #11555, and adds some more tests. Node is a different story, will tackle it separately.
Adds a test to ensure isolation scope is correctly applied to express spans, and also that console.log is correctly instrumented. We have related tests in E2E tests and in other places, but it doesn't hurt to have this explicitly here I guess...
Removes the `getCurrentHub` property from the `AsyncContextStrategy` interface and consequently also the property in the interface implementers.
This fixes the node SDK to make tracing without performance work as expected. This has three aspects: 1. Ensure we always have an url to check against `tracePropagationTargets`, even if the span is not sampled. 2. Ensure we use the correct propagation context always 3. Ensure we actually consistently handle ougoing http.client spans without an active transaction The fix for 2. was to have a fallback behavior if we don't have a scope linked to a context yet (we'll just use the current scope then). This is needed because we won't necessarily have an assigned scope yet at the time this is called, because at this point in time this is still a floating context. This also meant removing the fallback code to look at the remote span context (because we'll always have a scope now). The fix for 3. was to move this out of the http integration and do this in the sampler instead. The fix for 1. was a bit more elaborate, because we require different things to solve the http instrumentation and the fetch instrumentation. For fetch, we leverage our sampler, which can attach trace state that even unsampled spans will have. We attach the URL as trace state which we can read in the propagator, ensuring this works even for unsampled spans. Sadly, for http this is not sufficient, because our usage of `onlyIfParentForOutgoingRequest` lead to some spans being created outside of the sampler. By removing this option (see 3) and doing stuff in the sampler instead, we can ensure that all spans for http are handled properly. This also bumps our node-fetch otel instrumentation to [1.2.0](https://github.com/gas-buddy/opentelemetry-instrumentation-fetch-node/releases/tag/v1.2.0) which fixes headers for Node 21 (tests were actually failing without this!).
This is not supported since next 12, so we are good to remove this too: https://nextjs.org/docs/messages/webpack5
This is duplicated from the trace data, no need to send this anymore.
Lms24
reviewed
Apr 15, 2024
mydea
force-pushed
the
prepare-release/8.0.0-beta.0
branch
from
020f4d7
to
896d798
Compare
We don't actually need them anymore, as we've updated stuff I guess. This also finally fixes the problem that yarn always seemed to do some work even with the lockfile "up to date".
…n finished (#11600) Remove the now incorrect propagationContext resetting logic for pageload spans and add tests to cover this scenario.
AbhiPrasad
reviewed
Apr 15, 2024
| This is the first beta release of Sentry JavaScript SDK v8. With this release, there are no more planned breaking | ||
| changes for the v8 cycle. | ||
|
|
||
| Read the [in-depth migration guide](./MIGRATION.md) to find out how to address any breaking changes in your code. All |
There was a problem hiding this comment.
could we add notes about version changes and package removals here? https://github.com/getsentry/sentry-javascript/pull/11410/files#diff-06572a96a58dc510037d5efa622f9bec8519bc1beab13c9f251e97e657a9d4edR7
There was a problem hiding this comment.
will add the stuff you wrote there here too! 👍
Lms24
force-pushed
the
prepare-release/8.0.0-beta.0
branch
3 times, most recently
from
ed8d06c
to
409ca77
Compare
mydea
force-pushed
the
prepare-release/8.0.0-beta.0
branch
from
409ca77
to
a1b65e8
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
FIRST BETA INCOMING!