Important updates #2
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
…ng (and use it in the air traffic data example) (#8458) ### What It's some time nice to log some color information in multiple entities to make it easier to relate them visually. This PR adds a `rr.components.Color.from_str()` utility that does exactly that: generate a nice color randomly picked up based on the provided string. This PR also updates the air traffic data example so the barometric traces have matching colors with the map data. --------- Co-authored-by: Clement Rey <[email protected]>
#8468) ### Related <!-- Include links to any related issues/PRs in a bulleted list, for example: * Closes #1234 * Part of #1337 --> CI is failing on `nightly`: https://github.com/rerun-io/rerun/actions/runs/12335447188/job/34427482723 ### What Seems to be a mismatch between doc strings. <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. -->
## Related * Closes #8454 UI interaction was sometimes broken outside of graph views. This PR is the cure. * Reverts #8416 --------- Co-authored-by: Jochen Görtler <[email protected]>
### What @emilk and I discussed level-of-details on the graph. Here is an initial experiment. Overall, I'm not sure the performance improvements warrant this addition. Discussion + video: https://rerunio.slack.com/archives/C041NHU952S/p1734084741247629 <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. --> --------- Co-authored-by: Antoine Beyeler <[email protected]>
…8473) Wasn't exposed in earlier wgpu versions. Usually not a problem, but when it is it's usually too late ;)
### Related <!-- Include links to any related issues/PRs in a bulleted list, for example: * Closes #1234 * Part of #1337 --> * Closes #8437 * Closes #8442 ### What This implements: * Single-click on empty space to select view. * Double-click on node to select entire entity. Merging @emilk's recent changes (#8469 and #8457) seems to have fixed the flickering on selection too.
### Related * Closes #8263 ### What I couldn't stand the half-second delay when opening the options screen. Rerun needs to feel snappy! Gettings the ffmpeg version is now done on a background thread, showing s spinner until it is done. Unfortunately we still have to wait for ffmpeg when starting up an H.264 video on native. We could fix that by pre-warming the cache though 🤔 --------- Co-authored-by: Clement Rey <[email protected]> Co-authored-by: Andreas Reich <[email protected]>
### What This is not perfect, since Sorbet hasn't been formalized. This decodes the place-holder sorbet data we currently use in our query results, as well as some of the Rerun-chunk metadata. Eventually we should move this onto the rust-side of things, but as this is largely just metadata processing, doing it in python is not terrible.
…t invalid transforms (#8459) ### Related * Fixes #7868 ### What See title. In detail: * `RotationAxisAngle` will no longer be ignored when axis is non-normalizable * deprecation of `DisconnectedSpace` * slight fixes to codegen for handling this smoothly * lots of doc updates * wherever we had constants for identity, we now also have one for invalid * python didn't have any at all so far, but I didn't want to add this in this iteration since generally constants there are a bit tricky * we should have more of those generally --------- Co-authored-by: Jeremy Leibs <[email protected]>
…8369) ### Related * Fixes #8330 ### What Wasn't able to repro the issue in the first place though - tried fractional zoom and a contrast heavy background to no avail * [x] confirm this fixes it --------- Co-authored-by: Antoine Beyeler <[email protected]>
…/Show all` item (#8493) ### What ☝🏻 title. Formatting glitches were introduced in the "space view rename"
### Related <!-- Include links to any related issues/PRs in a bulleted list, for example: * Closes #1234 * Part of #1337 --> * Closes #8479 * Closes #5138 ### What This adds missing view interactions the following views: * Map view * Graph view * TextDocument view @Wumpf @abey79 I feel like the pattern repeats itself and it's easy to miss when writing new views. So I wonder if this could be unified. I see several options for this: * repurpose `handle_select_hover_drag_interactions` which only seems to be used for `DataResult` right now * Making each view return a `Result<(Response, ViewSystemExecutionError>` (instead of `()`) and handling it outside of the view. * Creating a function analogous to `handle_select_hover_drag_interactions`. <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. -->
### Related * Part of #8264 * Using emilk/egui#5438 Still missing: copy screenshot to clipboard Also fixes a crash when screenshotting graph views. ### Usage  ### Result 
### What Title. This was a small part of @gavrelina's feedback. <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. -->
…rly error on unsupported render targets (#8476) ### Related * Fixes #8466 * Related to #8475 ### What Improved & sharpend overall how we handle device tiers. Next steps that went too far in scope right now detailed in * #8475 ### Testing this shouldn't be all that problematic, but just in case let's be thorough: * [x] WebGL * [x] WebGPU * [x] Mac Metal * [x] Windows Vulkan * [x] Windows OpenGL * [ ] Linux Vulkan * [ ] Linux OpengL
…ead (#8467) ### What We have a foot-gun in our code: our file dialogs (via `rfd`) [are only allowed to be run from the main thread (at least on Mac)](https://docs.rs/rfd/latest/rfd/#macos-non-windowed-applications-async-and-threading). However, there is nothing stopping a user from accidentally calling these functions from a background thread, and if you test it on e.g. Linux, it may very well work. So this PR introduces a new crate `re_capabilities` and a new type `MainThreadToken`. Any function that uses `rfd` should require the `MainThreadToken` as an argument. The `MainThreadToken` is only allowed to be created in `fn main`, and since it is neither `Send` nor `Sync`, this guarantees at compile-time that any functions that take a `MainThreadToken` can only be called from the main thread. NOTE: I have no way to enforce that all uses of `rfd` also require `MainThreadToken` - we need to remember this ourselves, but I've made sure that all our _current_ uses of `rfd` require a `MainThreadToken`.
There are too many pitfalls that users can fall into because the current logging APIs are not designed with tagging in mind, and we have not yet communicated at all about any of this. This has to be done below the chunk level, directly at the descriptor level.
…n existing view in the viewport (#8431) ### Related * Closes #8266 * Related to #8267 ### What This PR makes it possible to drag one or more entities from the blueprint and streams tree to existing views in the viewport. This involves introducing and/or adjusting a whole bunch of semantics around selection and dragging. #### `DragAndDropFeedback` This PR introduced the notion of drag-and-drop feedback from the hovered ui element. Feedback may be either of: - **Ignore** (default): hovered element is uninterested in the type of drag payload, or in any payload at all, or is outright oblivious to all that drag-and-drop stuff. - **Reject**: hovered element is compatible with the _type_ of drag payload, but not its actual _content_. For example, a view might already contains the dragged entities. - **Accept**: hovered element is compatible with both the type and the content of the payload. A drop should only ever happen in the latter case. #### Payload visualisation The payload pill display is now adjusted based on the feedback, both its opacity (ranging from 50 to 80%) and its colour (grey or blue). The mouse cursor is also adjusted based on the feedback. #### Drop container visualisation This PR slightly adjust the look of the drop container visualisation: the blue frame is now 2px wide. Note that the drop container is _not_ necessarily the thing that's hovered by the mouse, see e.g. containers in the blueprint. #### Selection handling during drag-and-drop This PR slightly alter the current behaviour. Now: - dragging a selected item drags the entire selection - dragging an unselected item with `cmd` held adds that item to the selection, and drags the entire selection - dragging an unselected item drags that item, _without changing the selection_ (new) #### Entity-to-viewport-view drag semantics This is the original goal of this PR. - An existing view will accept a payload containing entities **IFF** any of these entities—or their children—is both visualisable and not already contained in that view. - An existing view will reject a payload containing entities **IFF** all of these entities are either non-visualisable or already contained. - An existing view will ignore a payload containing anything else. - When a drop succeeds: - The view will add an inclusive ("…/**") rule for each of the dropped entities that are both visualisable and not already included. - The view becomes selected. Emphasis on that last point. This subtle UX behaviour (courtesy of @gavrelina) makes the drop success and impact on the entity path filter more explicit. #### Theory of operation for drag-and-drop With this PR, a "framework" slowly starts to emerge. For now, it's mainly this bit of documentation: ```rust //! ## Theory of operation //! //! ### Setup //! //! A [`DragAndDropManager`] should be created at the start of the frame and made available to the //! entire UI code. //! //! //! ### Initiating a drag //! //! Any UI representation of an [`crate::Item`] may initiate a drag. //! [`crate::ViewerContext::handle_select_hover_drag_interactions`] will handle that automatically //! when passed `true` for its `draggable` argument. //! //! //! ### Reacting to a drag and accepting a drop //! //! This part of the process is more involved and typically includes the following steps: //! //! 1. When hovered, the receiving UI element should check for a compatible payload using //! [`egui::DragAndDrop::payload`] and matching one or more variants of the returned //! [`DragAndDropPayload`], if any. //! //! 2. If an acceptable payload type is being dragged, the UI element should provide appropriate //! visual feedback. This includes: //! - Calling [`DragAndDropManager::set_feedback`] with the appropriate feedback. //! - Drawing a frame around the target container with //! [`re_ui::DesignToken::drop_target_container_stroke`]. //! - Optionally provide more feedback, e.g., where exactly the payload will be inserted within //! the container. //! //! 3. If the mouse is released (using [`egui::PointerState::any_released`]), the payload must be //! actually transferred to the container and [`egui::DragAndDrop::clear_payload`] must be //! called. ``` ### TODO - [x] release checklist to check the above semantics https://github.com/user-attachments/assets/047c0d41-fead-424a-b673-b6cb1479d1fa
### What This PR builds proper scrollability to our modals, thereby fixing: - an ugly glitch with large add/remove entity modal - add view/container modal not scrollable with very small viewer window size #### Before <img width="625" alt="image" src="https://github.com/user-attachments/assets/cfe3d0cb-ac10-4bf5-84cb-4766589c5b41" /> #### After <img width="641" alt="image" src="https://github.com/user-attachments/assets/919cb62b-4934-429b-a41c-025f7c0cd9a9" /> <img width="951" alt="image" src="https://github.com/user-attachments/assets/e8eec505-a789-4285-b2c5-0c4906660fff" /> --------- Co-authored-by: Clement Rey <[email protected]>
### Related - fixes #8456 ### What ☝🏻 <img width="793" alt="image" src="https://github.com/user-attachments/assets/9ffea3a3-da2b-49dd-b5d8-0ce37491451f" />
Make sure we test everything that went into 0.30.0, while we wait for * lampsitter/egui_commonmark#69
* no hover at all * no highlight on non-text box nodes
### Related * Closes #8503. ### What This fixes setting the fallback value in the selection panel. It also deals with geometry that can't be drawn in a meaningful way because of that. <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. -->
* Closes #8494 Removes the highlighted rectangle around views that are hovered or selected. Why? Because we already have the tab title for this. Keeps it for containers, because they have no tab title. https://github.com/user-attachments/assets/bb407161-f3a8-4c66-bbd9-816f8dd28561
This allows the update APIs to now support multi-recording updates, and generally gives users more direct control over things like the column metadata, etc.
### Related * #7026 ### What Lots of considerations & (soft) implications in this hack, see comment. The important outcome here is that node hover now looks like this:  node selection:  and node entity selection (note that edges are still around:  also you can still browse the edges: 
The warning: ``` warning: /Users/hhip/src/rerun/crates/utils/re_capabilities/Cargo.toml: `default-features` is ignored for egui, since `default-features` was not specified for `workspace.dependencies.egui`, this could become a hard error in the future ```
This is a band-aid patch to address an issue with entity path filter which would result in the following behaviours: - the entity path filter would be perma-writen to the blueprint store when the view is selected (in some cases) - in such cases, this would break undo on user change (aka with entity drag-and-drop) The core issue is deeper and might cause panics with rust 1.81+. The TL;DR is that the current `EntityPathFilter` type is ambiguous as to whether its content has substitutions (aka $origin) applied or not. In particular `Eq` and `Ord` apply to unsubstituted, resp. substituted content (which can lead to the above panic). This should be further cleaned by having two structures, one unsubstituted and another substituted. --------- Co-authored-by: Clement Rey <[email protected]>
Makes the graph examples look a lil bit nicer and more compact. Note that this is a very recent thing that we can do this in Rust 🥳 Also, embedds `graph_directed` in the examples (this has been done directly on the release branch already) * [x] full check passed in order to confirm the example output is still the same (tested this locally but who knows)
### Related - #8522 ### What Explains how to avoid hitting an issue.
This PR enables the creation of GH releases (as draft and marked as pre-release) for RCs. Motivations for doing so: - test that part of the workflow, in particular that all binaries are correctly produced - test the binary using `cargo binstall` - these releases a single-click deletable from GH (if we want to do so) - GH releases marked as "pre-release" aren't displayed in the GH project page Pending questions: do we want that? what about alphas?
### Related * Closes #6830 ### What⚠️ Breaks tensors in .rrd files! In `TensorData` we used to have `shape: [TensorDimension]` with `struct TensorDimension { size: u64, name: Option<String> }`. Now `TensorData` has `shape: [u64], names: Option<[String]>` instead. So basically a AoS -> SoA change ### TODO * [x] Port Python * [x] Port C++ * [x] Document in migration guide * [x] Run `@rerun-bot full-check` --------- Co-authored-by: Antoine Beyeler <[email protected]>
### Related * fixes #8525 ### What Truncation now happens at egui test level, so we no longer need that manual truncation code. (This code is soon going to disappear anyways but at least it'll disappear in a good state 🤷🏻) 
Co-authored-by: Clement Rey <[email protected]> Co-authored-by: Andreas Reich <[email protected]> Co-authored-by: Katerina Gavrilo <[email protected]> Co-authored-by: Jochen Görtler <[email protected]>
### Related * #7485 * #8530 ### What Update MSRV to 1.81 --------- Co-authored-by: Jochen Görtler <[email protected]>
* `HybridResultsChunkIter::component` is now `HybridResultsChunkIter::component_slow`, in order to scare people off. * I have removed *a lot* of those, and will remove a few more in follow up PRs, but I need to add more iteration tools first.
* Follow-up to #8540 * introduce `iter_bool` * annihilate a bunch of `component_slow` using it
### What This mirrors the existing Recording API surface and query-construction, but dispatches the query through the the server. Without access to the schema, we are not yet able to support glob-expressions for content (rerun-io/dataplatform#105) Recommend viewing without whitespace: https://github.com/rerun-io/rerun/pull/8537/files?w=1
Co-authored-by: Andreas Reich <[email protected]>
This method makes no sense. It's a complete anti-pattern. The whole point of the Chunk level methods is to pay the cost of reflection/downcasting/deserialization once for the whole Chunk, this can never happen with the way this method is defined. I'm not sure what I was thinking back then. I likely wasn't. There is never a good reason to use this.
This is so that I can move `fn arrow_ui` (which depend on `SizeBytes`) into `re_ui`, as part of improved testing of it before migrating it to arrow1. And `re_byte_size` sounds nicer than `re_sizes_bytes`. Maybe we should rename the trait to `ByteSize` too. --------- Co-authored-by: Clement Rey <[email protected]>
### Related * Fixes #6817 * because, well, it got removed ### What As planned, removed `DisconnectedSpace` which was deprecated in 0.21 This makes `SpatialTopology` & `TransformContext` a little bit simpler, one thing less to take into account. Further simplifications should be possible now, but one thing at a time. --------- Co-authored-by: Clement Rey <[email protected]>
### What Title. Wonder if we should even show the forces section in the selection panel in this case—but that might be confusing for the user too. <!-- Make sure the PR title and labels are set to maximize their usefulness for the CHANGELOG, and our `git log`. If you have noticed any breaking changes, include them in the migration guide. We track various metrics at <https://build.rerun.io>. For maintainers: * To run all checks from `main`, comment on the PR with `@rerun-bot full-check`. * To deploy documentation changes immediately after merging this PR, add the `deploy docs` label. -->
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hi! Thanks for opening this pull request.
Because this is your first time contributing to this repository, make sure you've read our Contributor Guide and Code of Conduct.
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.
Related
What