src,lib: minor --debug-brk cleanup

[ofrobots]

Checklist

• tests and code linting passes
• documentation is changed or added
• the commit message follows commit guidelines

Affected core subsystem(s)

src,lib

Description of change

Minor cleanup of how --debug-brk works:

• We no longer need to use command line flags to expose the debug
  object.
• Do not depend on the existence of global.v8debug as a mechanism to
  determine if --debug-brk was specified. We may overwrite the debug
  listener with the dummy listener when --expose-debug-as=v8debug
  was otherwise present on the command line.
• We no longer need to set a dummy listener with --debug-brk.

This in anticipation of an upcoming PR with v8-inspector support.

R=@bnoordhuis
/cc @eugeneo

[ofrobots]

CI: https://ci.nodejs.org/job/node-test-pull-request/2516/

[cjihrig]

Is this related to #2546?

[eugeneo]

We believe this is a genuine bug, but we uncovered it while working on
inspector integration.

On Thu, May 5, 2016 at 2:00 PM Colin Ihrig notifications@github.com wrote:

Is this related to #2546#2546?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#6599 (comment)

[bnoordhuis]

LGTM but it would be even better if the property was deleted again before it's visible to user code.

[ofrobots]

@bnoordhuis That's a good idea. Please take another look. New CI: https://ci.nodejs.org/job/node-test-pull-request/2525/

[cjihrig]

LGTM

[bnoordhuis]

LGTM

[bnoordhuis]

Weird failure on the fedora23 bot, by the way: https://ci.nodejs.org/job/node-test-commit-linux/3256/nodes=fedora23/console

make[2]: Leaving directory '/home/iojs/build/workspace/node-test-commit-linux/nodes/fedora23/out'
ln -fs out/Release/node node
/bin/sh: line 1: ./node: No such file or directory
Makefile:148: recipe for target 'test/addons/.buildstamp' failed
make[1]: *** [test/addons/.buildstamp] Error 1
make[1]: Leaving directory '/home/iojs/build/workspace/node-test-commit-linux/nodes/fedora23'
Makefile:285: recipe for target 'run-ci' failed
make: *** [run-ci] Error 2
Build step 'Execute shell' marked build as failure

I don't know what's causing it but it looks unrelated.

[ofrobots]

Thanks. Landed as 4d4cfb2.

[MylesBorins]

@ofrobots lts?

[ofrobots]

I don't see a very strong benefit, or much harm from back-porting this to 4.x LTS. This should be back-portable, modulo the lib/module.js refactoring since 4.x. LMK if you want a PR w/ the backport.

[MylesBorins]

@ofrobots I'm moving this to don't land for now, please let me know if you would like to send a PR

[jez9999]

How am I supposed to programatically detect whether I'm in debug mode now if you don't expose global.v8debug? I tried checking process._debugWaitConnect but that is undefined too even when I pass --debug-brk=12345.

[bnoordhuis]

You could check with process.execArgv.some(s => /^--debug-brk=/.test(s)). I'm curious though, why does it matter to your code if debugging is on or not?

[jez9999]

@bnoordhuis Because if it's debugging, I need to change the argv for each new forked process so they don't have a debugging port clash with existing processes.

[bnoordhuis]

Are you using child_process.fork()? There was talk recently of extending the port shuffle mechanism in cluster.fork() to child_process.fork().

[jez9999]

Yeah, I am using that as I don't want processes to share ports. I don't know why the port shuffle wasn't extended, the principle is the same with child_process.

[jez9999]

In fact, is there any reason why i should use child_process instead of cluster? If the only difference is that cluster can share TCP ports, child_process seems to be redundant at this point. Kind of makes you wonder why they created a whole new core module for cluster.

[bnoordhuis]

With all due respect, but if you think that is the only difference between the two modules, you need to read the documentation (or their sources) more closely.

[jez9999]

From my reading of the docs that is basically the only difference. Also, this StackOverflow answer agrees with me. Can you point out any other differences?

[bnoordhuis]

Maybe you are simply being imprecise with language? You say 'module' but going by the link you posted, you really mean cluster.fork() vs. child_process.fork(). Is that correct?