meta(changelog): Update changelog for 10.0.0 #17271
Merged
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
Adds a test case with a client, server and server API request which checks: - that everything is correctly connected - that everything has parametrized routes (the current state of it - some are not yet parametrized) This is a visual representation of the test case (with a different param): <img width="502" height="197" alt="image" src="https://github.com/user-attachments/assets/e1a14c2b-2547-427b-964b-5c8fb6db49a2" /> Part of #16686
This allows us to avoid `any` here, and makes more sense overall IMHO.
This is IMHO not super helpful, we constantly disable it or add a noop comment for no gain.
[Gitflow] Merge master into develop
This PR * bumps all OpenTelemetry packages to v2 and equivalent (i.e. `0.202.0` for experimental and latest versions for all instrumentations). * removes our vendored express v5 instrumentation in favor of the OpenTelemetry instrumentation * adds polyfills for node utils to vercel-edge * removes the `@sentry/opentelemetry` OpenTelemetry v2 test dev-package as this is now covered by the package itself **Note**: Once this PR is merged, the `develop` branch officially becomes the `v10` branch and any releases for v9 have to happen on the `v9` branch.
) Co-authored-by: Charly Gomez <[email protected]>
`BaseClient` was deprecated in v9, use `Client` as a direct replacement.
`hasTracingEnabled` was deprecated in v9, use `hasSpansEnabled` as a direct replacement.
… sampling decisions (#17055) Add a couple of unit tests to explicitly test sampling behaviour of the core SDK's sampling logic when users pass in a `tracesSampler` function returning negative sampling values (i.e. `0`, `false`).
BREAKING CHANGE In #16901 we deprecated the `logger` export from `@sentry/core`. This PR removes it for the v10 release. --------- Co-authored-by: Andrei Borza <[email protected]>
This introduces a new AWS lambda layer that supports both ESM and CJS. Instead of bundling the whole SDK, we install the local NPM package and then prune all not strictly necessary files from `node_modules` by using `@vercel/nft` to keep the layer size as small as possible. closes #16876 closes #16883 closes #16886 closes #16879
…prior (#17091) CI for the 10.0.0-alpha.0 release is failing at the build step but it's unclear why. This hopefully reveals more info.
…ils (#17082) Sentry CLI's `cli.releases.uploadSourceMaps` method previously never rejected when the actual CLI binary execution exited with an error code. In CLI 2.49.0 (and 2.50.0) I added a new live mode `rejectOnError` which continues to pipe stdio to the process (the remix SDKs' upload script) but now also rejects on error. This PR 1. bumps Sentry CLI, 2. configures the CLI to actually reject now but 3. also catches the rejection and logs a message. I decided to still continue with the script because we should still delete source maps. Otherwise, we risk deploying them when users expect them to be deleted. (i.e. fail silently but correctly :D)
Just a minor cleanup from a TODO I left a while ago. We can safely remove this check because CLS isn't reported before FCP is reported.
The current way of building the AWS Lambda Layer is not working during releases because it tries to install @sentry/aws-serverless from the registry instead of the workspace. Temporarily disable it until we figure out how to fix it.
…ions (#17063) When we first shipped support for https://docs.sentry.io/product/explore/logs/ in the JavaScript SDK we gated the logging related options (`enableLogs`, `beforeSendLog`) behind `_experiments` namespace. See #15717 and #15814 for the PRs that made this change. For v10, we want to not make this experimental anymore, so we changed the logging options to become top-level options to `Sentry.init`.
Migration note for the changes in #17063
Looks like `example.com` has some kind of rate limit or random unavailability. Switching to `docs.sentry.io` resolved the flakes. Maybe we can replace all `example.com` uses in the codebase with a Sentry domain 🤔 Wdyt?
Forgot to disable the build in the bundle step.
Extracting astro-4 test cases (same as in astro-5) in a new PR while working on route parametrization. Part of #16686
We removed this in order to release a v10 alpha but the release process still expcets a built layer, thus failing. We'll wait for a proper layer build fix.
Small cleanup forward-port to `develop` (see #17095 (comment).) Co-authored-by: Andrei <[email protected]>
#17106) This fixes the AWS Layer release build which tried to fetch the new package versions from the NPM registry, which don't exist yet during the build. We now build a new `package.json` that uses `resolutions` to point all Sentry packages to their local version.
Next try. Use local file link here too to skip registry fetch
we didn't actually build the resolutions correctly, this should fix it
) Route Parametrization for Server Requests in Astro 5. The route information is gathered during build-time. During runtime, the route information is matched to use the parametrized route information during runtime Part of #16686
…to `0.2.2` (#17207) Although the way we pin this dep allows for patch versions, let's explicitly bump this because https://github.com/getsentry/sentry-javascript-node-native-stacktrace/releases/tag/0.2.2 includes a pretty nice fix for cross-platform timings. Thanks again to @matthew-nicholson-anrok for the contribution :)
Marks the captured error as `{ handled: false, type: 'fastify' }`
For v9, the changes were already reverted in this PR (#17179) to create a quick fix we can release soon. However, this removed some parametrization. This PR here not only fixed the problem with continuously writing to `globalThis` to share build-time data with the runtime (we don't do this anymore). The route parametrization now happens only during runtime, as we have access to the route segments at runtime with Astro v5. This adds a **little** performance overhead when compared with the previous approach (the route segments are now constructed during runtime) - but this is not an expensive operation. The `.find` method was used in the previous approach as well. Fixes #17179
Follow-up for this PR: #17168 Creating a shared function `flushIfServerless` as this functionality is used quite often in various SDKs.
Adds a Sentry-internal base interface for build-time options used in
Sentry's meta-framework SDKs (e.g., Next.js, Nuxt, SvelteKit).
SDKs should extend this interface to add framework-specific
configurations. To include bundler-specific options, this type can be
combined with one of the `Unstable[Bundler]PluginOptions` types, such as
`UnstableVitePluginOptions` or `UnstableWebpackPluginOptions` (also
added in this PR)
```typescript
import type { BuildTimeOptionsBase, UnstableVitePluginOptions } from '@sentry/core';
import type { SentryVitePluginOptions } from '@sentry/vite-plugin';
// Example of how a framework SDK would define its build-time options
type MyFrameworkBuildOptions =
BuildTimeOptionsBase &
UnstableVitePluginOptions<SentryVitePluginOptions> & {
// Framework-specific options can be added here
myFrameworkSpecificOption?: boolean;
};
```
I created some interfaces for better high-level organization:
- `SourceMapsOptions`
- `ReleaseOptions`
- `BundleSizeOptimizationsOptions`
---
Part of #14212
Fixed an issue where default exports in CommonJS were failing. Returning a new exports object was overriding some default export functionality. I removed some redundant code as well.
- Ref #17216 Because JavaScript can dynamically load code from text like this: ```ts const {dynamicFn} = await import(`data:application/javascript, export function dynamicFn() { throw new Error('Error from data-uri module'); };`); dynamicFn(); ``` You can get slack lines like this: ``` at dynamicFn (data:application/javascript,export function dynamicFn() { throw new Error('Error from data-uri module');};:1:38) ``` These can be huge, often base64 encoded strings. Currently we skip regex parsing on anything over 1k in length but that means we ignore the frame entirely. This PR: - Changes the stack parser so that if the line is over 1k, we still pass the first 1k for parsing - Adds initial check to the Node line parser that special cases truncated lines with data URIs in them
There are no more pre-releases planned, so we can already re-enable the layer.
This PR adds missing tool call attributes, we
add`gen_ai.response.tool_calls` attribute for OpenAI Chat Completion and
Responses API, supporting both streaming and non-streaming requests.
Core changes:
1. Request Side - Capture available tools:
- Extract tools + web_search_options from request params
- Set gen_ai.request.available_tools attribute
2. Response Side - Capture actual tool calls:
- Chat Completion: Extract from response.choices[].message.tool_calls
- Responses API: Extract from response.output[] (filter type ===
'function_call')
- Set gen_ai.response.tool_calls attribute for both
3. Streaming Support (in streaming.ts):
- Accumulation of tool calls during streaming
- Respects recordOutputs privacy setting
…fault (#17220) Flushing on feedback widget opening is better than on submit as you get more relevant replay context, so we are going to make this the default behavior and remove the configuration option. This is deprecated in #17219 --------- Co-authored-by: Charly Gomez <[email protected]>
andreiborza
requested review from
Lms24,
AbhiPrasad,
chargome,
msonnb,
s1gr1d and
RulaKhaled
AbhiPrasad
approved these changes
Jul 31, 2025
s1gr1d
approved these changes
Jul 31, 2025
size-limit report 📦
|
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.
🤞 🤞 🤞