* build: add watch/compile tasks for CLI
I spent time this morning working on the 'developer experience' of the
CLI in vscode, mainly getting the CLI to cross-compile chasing our
initial idea of having it auto-build in a devcontainer.
After some effort and disabling tunnels connections (to avoid having to
pull in OpenSSL which is a huge pain to cross compile), I was able to
get it to cross-compile from Linux to Windows, using the mingw linker.
I could probably figure out how to get macOS working as well with more
effort. However, I'm not a big fan of this, effectively it's one more
'platform' of build we need to support and test.
I think a better approach is downloading the latest compiled CLI from
the update server instead, as needed. That's what this PR does. It just
places the CLI where it _would_ normally get compiled to by cargo; so
far we don't need to do anything special outside of that.
A notice is shown to users if this fallback happens.
* update from review
* Revert "Have English language return default result (#175157)"
This reverts commit 3c9f409888.
* Revert "Fix locale and language handling (#174779)"
This reverts commit 4e32196835.
* Revert "Add en as fallback osLocale (#175039)"
This reverts commit 7b6b6869e8.
* Revert "Ensuring locale is set to the OS locale and language is set to the Application Language (#174733)"
This reverts commit ecf479e662.
* Implement titlebar correctly using platformLocale
* Add more logging and perf markers around resolving the connection token and the socket factory
* set `exposeFunction` earlier
* bla windows
* also expose function for unit tests beofre opening
Co-authored-by: Benjamin Pasero <benjamin.pasero@microsoft.com>
* ipc: use vql for uint types
On the plane I was reverse-engineering ipc.ts to implement it in Rust
and see if we could have a "service mode" for the CLI that we could
interact with like any other vscode process.
In doing so, I noticed that numbers in the protocol--which are used at
least twice in the message header and ID--were encoded as JSON. I was
curious what benefits we'd get from encoding them as variable-length
integers instead.
It makes the message shorter, as expected. Encode/decode time are very,
very slightly lower. I'm not sure it's worth the extra complexity, but
I have included it here for your consideration.
* fixup tests
* Highlight label should not create extra empty dom nodes
I noticed that the `HighlightedLabel` class creates extra `span` elements for text ranges. These should not be needed. Using text children directly should be faster for creation and also reduce the number of nodes in the document
I also related the conditional spread with a longer version that uses a simple call to push. This is worth doing since `HighlightedLabel` is so widely used in the editor
* Update tests
* Update smoke test selector
* Update css
* Remove vscode-notebook-tests in favor of an .ipynb in vscode-smoketest-express
* Update build folder
* Add build task to correct platform
* Build for smoke tests on other platforms
* Fix repo url and remove comment
* Just -media?
* Update darwin/win32 as well
