optimize no-cycle rule using strongly connected components #2998
Conversation
…ErrorMessagePath` for faster error messages
|
This fails some tests. I need help with dealing with these edge-cases. |
|
New and removed dependencies detected. Learn more about Socket for GitHub ↗︎
|
|
👍 Dependency issues cleared. Learn more about Socket for GitHub ↗︎ This PR previously contained dependency changes with security issues that have been resolved, removed, or ignored. |
|
Why is |
|
Good question. I think it's that this is the first new PR it's been run on since i installed it? |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #2998 +/- ##
===========================================
- Coverage 95.66% 85.41% -10.26%
===========================================
Files 78 79 +1
Lines 3275 3325 +50
Branches 1150 1159 +9
===========================================
- Hits 3133 2840 -293
- Misses 142 485 +343 ☔ View full report in Codecov by Sentry. |
| @@ -62,6 +68,8 @@ module.exports = { | |||
| context, | |||
| ); | |||
|
|
|||
| const scc = StronglyConnectedComponentsBuilder.get(myPath, context); | |||
There was a problem hiding this comment.
Can this row and line 114-117 move after line 119?
when skipErrorMessagePath is false, there is no need to do scc check.
There was a problem hiding this comment.
the scc check is how we know if there's cycles or not.
|
|
||
| export default class StronglyConnectedComponentsBuilder { | ||
| static clearCache() { | ||
| cache = new Map(); |
There was a problem hiding this comment.
| cache = new Map(); | |
| cache.clear(); |
this might be less GC intensive? i don't feel strongly tho
There was a problem hiding this comment.
Maybe. clearCache is currently used only by tests, so I think it doesn't matter.
ljharb
force-pushed
the
leverage-scc-in-no-cycle
branch
from
f64a123
to
a47b5b9
Compare
Lets make the
no-cyclerule faster by not running many unnecessary BFSes.For each dependency graph (aka
ExportMap) we can run Tarjan's SCC once (which is a derivative of DFS = O(n))That saves us a lot of work because we run a linear-complexity algorithm once, as opposed to for each linted file (which turned us O(n^2))
#2937