Make WordPress Core

Opened 3 years ago

Closed 3 years ago

#45455 closed defect (bug) (duplicate)

Webpack "stuck" at old package versions

Reported by: azaozz Owned by:
Milestone: Priority: normal
Severity: normal Version: 5.0
Component: Build/Test Tools Keywords:
Focuses: Cc:


There are large differences between /dist on a local build of the 5.0 branch and
https://build.trac.wordpress.org/browser/branches/5.0/wp-includes/js/dist/. Seems the local build is stuck at older versions of the packages.

Deleting node-modules and reinstalling them seems to fix the issue, but that's not practical to do all the time :)

Also, this could be Windows related.

Change History (9)

#1 @azaozz
3 years ago

Tested with local built of RC-2 rev. [43958] comparing to https://build.trac.wordpress.org/browser/branches/5.0 on rev. 43790 (build after the same commit).

Last edited 3 years ago by azaozz (previous) (diff)

#2 @pento
3 years ago

  • Milestone changed from 5.0.1 to 5.0.2

#3 @pento
3 years ago

  • Milestone changed from 5.0.2 to 5.0.3

This ticket was mentioned in Slack in #core by audrasjb. View the logs.

3 years ago

#5 @audrasjb
3 years ago

  • Milestone changed from 5.0.3 to 5.1


Per today's bug scrub, let's address this ticket in 5.1.

#6 @netweb
3 years ago

Related: #45815 Build repository has unused '/js' and '/styles' directories

#7 @atimmer
3 years ago

My understanding is that package-lock.json should handle this. If it doesn't then either the package-lock.json is incorrect or npm is not working correctly. As you get the correct dependencies when deleting node_modules I am tending towards npm not working correctly. Maybe there is a flag that can force npm to check if everything in node_modules is still up to date?

#8 @netweb
3 years ago

It might be time to drop the ~ and ^ semver ranges from package.json per #45149

#9 @pento
3 years ago

  • Milestone 5.1 deleted
  • Resolution set to duplicate
  • Status changed from new to closed

This sounds like npm allowing looser semver ranges than it really needs to.

Going to close this in favour of #45149.

Note: See TracTickets for help on using tickets.