-
Notifications
You must be signed in to change notification settings - Fork 1.3k
Migrate to a better file watcher #636
Comments
+1, I basically chose gaze because it's widely used in gulp and seems to work there. But I'm all for a change. |
Gulp is switching to chokidar ;) gulpjs/glob-watcher#13 |
👍 Lets make it a post-v2 (stable) change. BTW, nice name for a watcher library: Chokidar (watchman in my lang) :p |
👍 |
FWIW, I've worked a lot with gaze, chokidar and sane. sane is the newest and least well known, but it is by far the most reliable in my experience. And I think it has the most sensible API. Worth a look. |
@callumlocke, thanks for the suggestion. Will take a look. |
👍 would really like this! |
👍 Gaze is using ancient deprecated dependencies. |
Some help is needed to fix the watcher tests, we had to disable them because the way the CLI was called didn't really work. |
+1 to swtich to chokidar, or at least update the gaze dependency, as the current version used by node-sass contains very outdated deps:
|
Why does node-sass even include a Why not improve the focus of this package and remove the |
@cody-greene I think there is some value in having it, but agree it should be separated out from this module, eg sass vs sass-cli. Lots of other modules have made this change recently and it works well imo |
@callumlocke Yes there is value in the feature. I can understand a few people are definitely using it. But I meant to suggest that the value provided is vastly overshadowed by these other issues. |
Can I install a "chokidar version" of node-sass now 😍?! node-sass can't watch files for a long time, and do not know why, now I'm using |
Can I open a PR to switch to Chokidar? The use of Gaze makes my CPU burn at 150-300% when using with docker for mac, with Chokidar my CPU runs at a normal 6-10%. I will start working on this as it really prevents me from using node-sass, or docker-for-mac. What do you need to see in my PR to get it merged? Edit: I see a PR already open to do this #844 |
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (sass#636) becomes a viable option. Closes sass#1896 Fixes sass#1891
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (sass#636) becomes a viable option. This PR not only massively increases our test coverage for the watcher but also address a bunch of known edge cases i.e. orphaned imports when a files is deleted. Closes sass#1896 Fixes sass#1891
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (sass#636) becomes a viable option. This PR not only massively increases our test coverage for the watcher but also address a bunch of known edge cases i.e. orphaned imports when a files is deleted. Closes sass#1896 Fixes sass#1891
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (sass#636) becomes a viable option. This PR not only massively increases our test coverage for the watcher but also address a bunch of known edge cases i.e. orphaned imports when a files is deleted. Closes sass#1896 Fixes sass#1891
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (sass#636) becomes a viable option. This PR not only massively increases our test coverage for the watcher but also address a bunch of known edge cases i.e. orphaned imports when a files is deleted. Closes sass#1896 Fixes sass#1891
This logic is all tightly coupled in the bin. This logic has been notoriously impossible to test due to weird fs timing issues. By extracting the actual logic we're now able to test it in isolation. With this in place replacing Gaze (#636) becomes a viable option. This PR not only massively increases our test coverage for the watcher but also address a bunch of known edge cases i.e. orphaned imports when a files is deleted. Closes #1896 Fixes #1891
Hey. Your file watcher (gaze) has 53 opened issues. Many bugs on various platforms.
Can we make a pull request that migrates the watching to chokidar? Chokidar is ultra-stable and is used in many popular projects.
The text was updated successfully, but these errors were encountered: