Summary:
Addressing code review feedback from previous 3 diffs.
* Don't return draft when using immer
* Move functions out of component to module level
* Make executeShell async and simplify callsites
* Add onStopMonitor to cleanup function
* Add uniform spacing by adding Layout.Container around titles
* Disable searchbar on both tables
* Use Toolbar
Reviewed By: mweststrate
Differential Revision: D28091565
fbshipit-source-id: 533d2491e6f48a9eaaa64b1a6cf76eea2e7189a5
Summary: Finishes off by converting the rest of the UI.
Reviewed By: mweststrate
Differential Revision: D28061468
fbshipit-source-id: 8d87d3344fe3937470e566d8c425efc97f38333c
Summary:
Half way through sandy conversion I ended up redoing the task scheduling in this plugin because it was sketchy and unreliable.
Previously, it was using setInterval to schedule a task every 500ms, which would query a lot of data from the device. The problem was that if these tasks took >500ms (which they did for me at least), you'd get overlapping ones, causing havoc.
Now it uses setTimeout instead, promisifies everything, and only ever schedules the next timeout when the current task has finished.
Also fixed a race condition where you could hit stop, but not actually stop the recurring tasks... and then there was no way to stop them because it was already "stopped".
Reviewed By: mweststrate
Differential Revision: D28061469
fbshipit-source-id: 2bb8aba7ddf37ba1fdbdc52c95ec258cae96f509
Summary:
Starts converting the plugin by moving to new model and switching some of the UI components.
No logic changes here, it's just shifting everything over.
Reviewed By: mweststrate
Differential Revision: D28028052
fbshipit-source-id: 7444eab77d9dba49261a70887ee26b8d120e97a6
Summary: This fixes an earlier reported issue with Messages plugin, if no column widths are set values can jump around per row
Reviewed By: passy
Differential Revision: D28090807
fbshipit-source-id: be124b94f507584cf177710816035cd280a5ef01
Summary:
4.10.2 had an fburl.com link in it when I should have used an fb.me link :( I'm sorry
cc mweststrate
Pull Request resolved: https://github.com/facebook/flipper/pull/2244
Reviewed By: passy
Differential Revision: D28054736
Pulled By: mweststrate
fbshipit-source-id: 0330f884e5d82cc669232aee3d8439b0f51240bd
Summary: I found a bug after migration to Sandy. It is now not possible to add queries to favorites. This diff fixes the issue.
Reviewed By: mweststrate
Differential Revision: D28066435
fbshipit-source-id: 129a62dab9521160f8957abba70e5c7a6e609839
Summary: The main change is the move to DataTable here.
Reviewed By: mweststrate
Differential Revision: D28006097
fbshipit-source-id: 7564276a1177a7835612db08857862cb81942bce
Summary: Some non-UI tests. Couldn't quite figure out how to make the renderer aware of changes in the old ManagedTable. Given that I've got a task for migrating, I didn't think it was worth digging into it more.
Reviewed By: mweststrate
Differential Revision: D28001651
fbshipit-source-id: 034b3d71bff2513c946b9f84525f3344b7879df9
Summary:
Protobuf based APIs are becoming more common (i.e. gRPC) but are difficult to inspect. Unlike plain text data formats (JSON), Protobuf calls transmit binary data requiring the format to be known ahead of time, making ad-hoc inspection impossible. This PR allows for those format definitions (messages in protobuf terminology) to be transmitted from the client to the network inspector plugin. These definitions are then imported into ProtobufJS which enables the binary data transmitted to be inspected as easily as JSON data.
See Retrofit PR in https://github.com/facebook/flipper/pull/2084
## Changelog
* Add ProtobufJS library to network plugin
* New `ProtobufFormatter` UI in `RequestDetails`
* `ProtobufDefinitionsRepository` to cache and load protobuf defintions
* `addProtobufDefinitions` call in the Android network plugin
Pull Request resolved: https://github.com/facebook/flipper/pull/2080
Test Plan: 
Reviewed By: mweststrate
Differential Revision: D27507451
Pulled By: passy
fbshipit-source-id: 586d891b74f2b17d28fe7a2a99074da755851f38
Summary:
Reverts D27764688 (7bd4f80c25) due to a bug with "Invalid hook call." being erroneously reported. We will upgrade again after that bug is resolved.
Changelog:
[Internal]
Reviewed By: fkgozali
Differential Revision: D27813660
fbshipit-source-id: 84a12f19cf1bb7e8aebef0da3ff6f7022c391d3e
Summary:
Layer selection is pretty easy to miss, as reported in for example: https://fb.workplace.com/groups/flippersupport/permalink/1098169193997071/
Moved the layer selection to the top of the view and gave it some highlighting + dynamic height. The section is no longer collapsible.
Changelog: [Layout] Make the layer selection more prominent
Reviewed By: priteshrnandgaonkar
Differential Revision: D27708650
fbshipit-source-id: c86a55c3a20794aee86e64b6766b2ca4dd6b563f
Summary:
This diff fixes some more perf bottlenecks in the layout inspector (see the diffs earlier in the stack for the total picture). Mostly:
1. Pass down stable refs from the root for callbacks and configuration
2. Remove the deep-equality check in the sidebar section rendering, which has a pretty significant constant overhead, especially if the selection didn't change
3. If the selection changes, the correct semantics is to reset the sidebar rather than trying to reconcile the elements. (A consequence of this is that Panel collapse state isn't preserved atm after changing selection, will address that in a later diff)
This reduces average render time for sidebar from ~20 to ~2 ms.
Reviewed By: priteshrnandgaonkar
Differential Revision: D27677353
fbshipit-source-id: ba183b7e3d778c0b3c8e7ca0d51535ce99a097ca
Summary: Sidebar extensions were unkeyed, which make them render inefficiently. I cleaned the api a bit here by making the extensions more idiomatic; they are now components rather then functions, so that they have their own render cycle, state, etc. They are memo-ed now as well, so that they don't have to re-render if the selected item doesn't change.
Reviewed By: nikoant
Differential Revision: D27685980
fbshipit-source-id: b133bc42061b3b8cf971792f5818810ecb80e3ea
Summary:
Changelog: [Layout] Addressed several performance issues in the layout plugin
This diff and a few of the next stuff fix some performance issues in the Layout plugin. This diff fixes an issue where computing the context menu will cause all rows to render at all times, make the responiveness of the plugin quite slugish.
The fix in this case is to build up the context menu lazily, and pass a stable ref to the function through the tree, rather than a new menu every time the root component renders.
The changes in this diff and the next ones in total reduces the time (in prod builds) to draw a frame from ~200ms to ~5ms.
Reviewed By: cekkaewnumchai
Differential Revision: D27685983
fbshipit-source-id: a48b2ce2cdd1db31bb13122924617cbc3b6c198a
Summary:
This diff moves the core of ElementsInspector to flipper-plugin and decouples it from legacy design system and Electron, without any significant improvements or API changes yet, which will follow later.
Colors and docs will be added later in this stack.
Reviewed By: passy
Differential Revision: D27660300
fbshipit-source-id: 96abfa3b3174fa852cf04ae119c23c3d629fee74
Summary:
Pull Request resolved: https://github.com/facebook/flipper/pull/2172
New script which runs "tsc" for all plugins, receives list of errors and then checks which plugins are affected. It works for shared libs too, e.g. if there is an error in a shared library, then all plugins dependant on it will be counted as affected. For convenience, script saves list of errors affecting each plugin to "tsc-errors.log" in plugin folder.
This script will be used for automatic type-checking plugins against current "stable" and "insiders" versions of Flipper.
An alternative to this implementation would be to simply run "tsc" for each plugin individually, but such implementation takes a lot of time (5+ sec per plugin) and so cannot be effectively used on diffs.
Reviewed By: mweststrate
Differential Revision: D27499765
fbshipit-source-id: fcbbfc94a13e6c7c5beff0c889a929f84c41b2dd
Summary: This is a guard to ensure all Flipper built-in packages are declared as peer dependencies. It also removes all of them from nested node_modules after installation to be 100% sure they always loaded from the root folder, because e.g. react can be still installed as a transitive dependency even it is not declared as dependency directly.
Reviewed By: passy
Differential Revision: D27040267
fbshipit-source-id: 1e315a6b280b36ab20778ee261aa386b51d9f964
Summary:
Here I'm changing plugin repository structure to allow re-using of shared packages between both public and fb-internal plugins, and to ensure that public plugins has their own yarn.lock as this will be required to implement reproducible jobs checking plugin compatibility with released flipper versions.
Please note that there are a lot of moved files in this diff, make sure to click "Expand all" to see all that actually changed (there are not much of them actually).
New proposed structure for plugin packages:
```
- root
- node_modules - modules included into Flipper: flipper, flipper-plugin, react, antd, emotion
-- plugins
--- node_modules - modules used by both public and fb-internal plugins (shared libs will be linked here, see D27034936)
--- public
---- node_modules - modules used by public plugins
---- pluginA
----- node_modules - modules used by plugin A exclusively
---- pluginB
----- node_modules - modules used by plugin B exclusively
--- fb
---- node_modules - modules used by fb-internal plugins
---- pluginC
----- node_modules - modules used by plugin C exclusively
---- pluginD
----- node_modules - modules used by plugin D exclusively
```
I've moved all public plugins under dir "plugins/public" and excluded them from root yarn workspaces. Instead, they will have their own yarn workspaces config and yarn.lock and they will use flipper modules as peer dependencies.
Reviewed By: mweststrate
Differential Revision: D27034108
fbshipit-source-id: c2310e3c5bfe7526033f51b46c0ae40199fd7586
