Leverage node-glob stats cache #237
Conversation
Implemented `overwrite` option in `dest()`
Fix mode comparison
this relies on the recent breaking changes to glob-watcher being published to npm as 2.0
switch to glob-watcher 2.0
It has been shown in gulpjs#62 this bit can be platform specific and is unreliable, testing fs.chmod is not called will be enough.
Remove realMode equality test on special bit test
export `filterSince` to `vinyl-filter-since`
specifying the type and URL is deprecated: https://docs.npmjs.com/files/package.json#license http://npm1k.org/
update license attribute
Update vinyl to 0.5.x
|
This needs to be rebased (including the new options resolution), but I'm not sure how. My first thought is that |
|
@erikkemperman I've been running some benchmarks with your |
|
@phated I have a hunch the performance degradation is elsewhere, and might be sufficiently bad that it defeats measuring the impact of using this cache. Looking into that now... |
|
@erikkemperman I agree that the main degradation is elsewhere, but this actually reduces the performance more (on top of that). Hopefully we can figure it out. |
…nts instead of filesystem times (fixes gulpjs#208)
erikkemperman
force-pushed
the
use-glob-cache
branch
from
ac808a0
to
4b76530
Compare
|
Rebased, would depend on (something like) gulpjs/resolve-options#3 |
|
This was closed due to me rebasing master. Please resubmit it to the updated codebase. |
|
@erikkemperman just noticed this was never re-opened after my rebase. Still have interest in it? |
|
@phated Ah, thanks for reminding me! Yeah, I looked into it briefly, trying to figure out exactly in which cases node-glob populates the stat cache. Turns out it's only for patterns that don't have any wildcards in the last path component, e.g. I'd be happy to rebase, but would need something like gulpjs/resolve-options#3 in place, if I remember correctly. |
@phated As discussed elsewhere, a PR that might serve as discussion piece on the merits and perils of node-glob's cache(s).