Update lerna

[nicolo-ribaudo]

Q	A
Tests Added + Pass?	Yes
Documentation PR Link
Any Dependency Changes?	Yes
License	MIT

[nicolo-ribaudo]

Version 14.2.1 generates an error:

TypeError: Cannot delete property '--' of #<Object>
    at Object.Yargs.self._copyDoubleDash (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:1196:5)
    at Object.parseArgs [as _parseArgs] (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:1097:60)
    at Object.parse (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:578:25)
    at main (/home/nicolo/Documenti/dev/babel/babel/node_modules/lerna/index.js:44:6)
    at Object.<anonymous> (/home/nicolo/Documenti/dev/babel/babel/node_modules/lerna/cli.js:11:15)
    at Module._compile (internal/modules/cjs/loader.js:689:30)
    at Object.Module._extensions..js (internal/modules/cjs/loader.js:700:10)
    at Module.load (internal/modules/cjs/loader.js:599:32)
    at tryModuleLoad (internal/modules/cjs/loader.js:538:12)
    at Function.Module._load (internal/modules/cjs/loader.js:530:3)

because of

babel/Makefile

Line 240 in 6ba1131

yarn lerna bootstrap -- -- --ignore-engines

cc @bcoe (author of the commit in v14.2.1 - yargs/yargs@e78e76e) is this a known issue fixed in v15?

[JLHwung]

Could we work around it by

node $(npm bin)/lerna bootstrap -- --ignore-engines

[bcoe]

@nicolo-ribaudo something outside of yargs, in lerna is calling Object.seal() on the argv object, before the parse is finished (there's a two pass parse that temporarily stores positional arguments in -- and then merges them into the final parse).

[nicolo-ribaudo]

@JLHwung Nope:

node node_modules/.bin/lerna bootstrap -- --ignore-engines
/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:1184
      else throw err
           ^

TypeError: Cannot delete property '--' of #<Object>
    at Object.Yargs.self._copyDoubleDash (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:1196:5)
    at Object.parseArgs [as _parseArgs] (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:1097:60)
    at Object.parse (/home/nicolo/Documenti/dev/babel/babel/node_modules/@lerna/cli/node_modules/yargs/yargs.js:578:25)
    at main (/home/nicolo/Documenti/dev/babel/babel/node_modules/lerna/index.js:44:6)
    at Object.<anonymous> (/home/nicolo/Documenti/dev/babel/babel/node_modules/lerna/cli.js:11:15)
    at Module._compile (internal/modules/cjs/loader.js:689:30)
    at Object.Module._extensions..js (internal/modules/cjs/loader.js:700:10)
    at Module.load (internal/modules/cjs/loader.js:599:32)
    at tryModuleLoad (internal/modules/cjs/loader.js:538:12)
    at Function.Module._load (internal/modules/cjs/loader.js:530:3)
make: *** [Makefile:240: lerna-bootstrap] Error 1

@bcoe Thanks! I'll fix it in lerna then.

[bcoe]

@nicolo-ribaudo could you try commenting out this line?

https://github.com/lerna/lerna/blame/2c487fe5f612739970eb08026b2bf7da82a32b66/core/command/index.js#L99

I haven't quite figured out how to make a reproduction in yargs; If we can figure out a reproduction, I'd be happy to add a try/catch, and just skip the delete step for this edge case.

This does not fail, like I would hope:

  // see: https://github.com/babel/babel/pull/10733
  it('does not fail if Object.seal() has been called on argv', () => {
    Object.seal(process.argv)
    const argv = yargs()
      .command('cmd <version>', 'a command', () => {}, (argv) => {
        Object.freeze(argv)
      }).argv
    console.info(argv);
  })

[evocateur]

Ack, sorry folks. I don't even remember why I cared about freezing the argv. I can remove it, as it serves basically no purpose.

[bcoe]

@evocateur I think you should be able to freeze the object once it's been passed to a command ... so it feels like yargs stepping on your toes as well.

I was thinking this morning, perhaps we pass a shallow clone of the object to the command rather than the actual argv.

[evocateur]

I'm fixing it in lerna/lerna@8ca85a4 by deep cloning argv before I mutate it. Releasing momentarily...

[evocateur]

lerna v3.18.5 published with the aforementioned fix

[existentialism]

thanks @evocateur!