* Mangle exported functions
For #180461
This prototype tries to mangle exported functions, saving a further 440kb from the bundle size
* Fix missing call
* Also try mangling top level exported consts too
* Fixing errors
* Don't run on build files
* Skip a few more manglings and revert change to namespace
* Skip a few more monaco files
* Also mangle consts that shadow types
This increases savings up to 3325
* Also mangle exported classes
* Skip mangling more localization functions for now
* Opt out pfs
* Update build script
* Run find locations task in parallel
This should speed up compile
* Cleanup before close
* Limit workers to avoid hitting memory limit
* Limit pool size
* Skip one more mangling
* Exclude entrypoints from mangling
* Try to fix web build and clean up code
* Exempt a few more projects
* Exempt another file
* Also exempt html
* Skip mangling ext entrypoints
* Use prefix that can't be confused with rpc calls
* Fix max call stack error
* Switch prefixes
* Don't mangle ambient declarations
* Use correct way of checking modifier flags
* Workaround getCombinedModifierFlags not doing what I'd expect
Maybe needs the checker to be enabled too? Just check parent chain instead for now
* Clean up code and add logic showing how enum mangling could work
* Remove a few more skipMangles
Use entrypoints instead
* Fix entrypoint name
* build: don't apply cli 'patches' for merge check
This also needed modules installed. But with the new toml patches, the 'merge' will never fail outright, so there's no need to do a separate merge check for it imo
* reduce duplication in pipelines, cache node_modules
Currently the Cargo.toml patches are git patches, but these can break
easily when surrounding lines are changed. Instead this lets us make
"Cargo.patch.toml" which gets a structured merge.
It did not normalize to forward slashes, so the empty sourcemap was
created with a source called `foo\\bar.js`. Then the build process was
adding a new source `foo/bar.js` since it didn't see it already in the
sourcemap, but `foo\\bar.js` was later normalized. This normalized
version came first in the mapping, but didn't have any mapped locations,
so breakpoints never got set.
* Add settings for fixed-width tabs
This is meant at least partially to address #40290 and
is a continuation of the unfinished work from #40750.
* Only apply fixed width when the setting is on
* Implement chrome-like tab width behavior
Tabs shrink uniformly (down to a limit) but stay fixed-width
when the mouse is over the tab bar.
* Rename width setting to max width
* Make the ifs more readable
* Have event handlers only if the option is set
* 🎨
* Handle sizing artifacts present with wrapping tabs
To achieve this, it's best to remove the transition delay.
* Rename setting to apply only for fixed-sized tabs
* Set default fixed max tab width to 160px
* Minor code tweaks from review comments
* formatting
Co-authored-by: Benjamin Pasero <benjamin.pasero@gmail.com>
* Use disposable listeners
* Remove redundant check for last-in-row
* Apply fade gradient from shrink tabSizing to fixed
* Trying to make the code cleaner and understandable
* Remove transition
On advice of @bpasero, removed transition because
the editor doesn't really use transition that much.
* some cleanup
* fix typo
* Simplify workaround for the overflow issue
* 💄
* 💄
* Let the tabs fix their width before closing the editor
* 💄
---------
Co-authored-by: Benjamin Pasero <benjamin.pasero@microsoft.com>
Co-authored-by: Benjamin Pasero <benjamin.pasero@gmail.com>
* innosetup to start/stop tunnel service
* better messages for tunnel still running
* fix error
* more logs
* add '/log'
* update
* fix
* update
* add kill & timeout wait
* stop service in PrepareToInstall
* non-background: check for tunnel mutex when wizard starts
* polish tunnel mutex names, prompt if tunnel is still running
