-
Notifications
You must be signed in to change notification settings - Fork 29.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
proposal: require('perf') as API for metric based data #6022
Comments
Are you proposing a general performance API (not necessarily related to the current process) or an API as it directly relates to the current process only? |
Things relating to the current process should probably be on If it's across you're OS, wouldn't |
IMO the data that Or would you like to split the (Just as a side note: A long, long time ago, I proposed a PR with |
It seems to me that things within the "perf" scope would apply to the current process so it's not too hard to justify extending the If we were to introduce a new core module we'd either need to come up with a name that doesn't introduce a conflict with userland or negotiate with the owner of the name we want to take ownership of it to core, currently for "perf" that's https:/reqshark/perf |
Yeah, -1 on a new module for this. |
Okay. I then I am going to continue this in the scope of process. I just wanted to avoid sementaic collisions, for when you are measuring hardware. But maybe this then could be put into others APIs from the Side note on |
I don't believe this needs to stay open. Closing but we can reopen if necessary |
As mentioned on #5565 (cc @mscdex, @evanlucas) I would like to propose a subsystem that would hold perf related data. Currently those are scattered mostly in
process
, where they might not fit.Also I am researching C APIs for Linux perf_events / Windows PDH / BSD Dtrace (in C) that could (theoretically) be easily added to the debugger e.g. for stack traces beyond
js
.libuv
already provides APIs likeuv_getrusage
that can be easily extended into node.Since the C APIs offer a multitude of information the API could potentially be to bloated for
process
and could potentially includethread
related or other data, that don't fit the semantic context.Action Items
libuv
to fill up some windows metrics of the unix namedgetrusage
(first stab was here util: extends uv_getrusage by majflt on NT and other NT counters libuv/libuv#787). Afterwards I would send in a PR to node to exposeuv_getrusage
. The question would be whether it should go into a new subsystem, intoprocess
or intoprocess
and maybe later into a new subsystem.The text was updated successfully, but these errors were encountered: