fs: two minor optimizations #14055

BridgeAR · 2017-07-03T13:29:22Z

If the function in tryStatSync threw, it would never reach that code path to return any value.
Only use try catch if necessary. lchmodSync does not need the second try catch in most cases.

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
tests and/or benchmarks are included
documentation is changed or added
commit message follows commit guidelines

Affected core subsystem(s)

fs

TimothyGu · 2017-07-04T08:31:51Z

lib/fs.js

@@ -552,10 +551,11 @@ fs.readFileSync = function(path, options) {
  var isUserFd = isFd(path); // file descriptor ownership
  var fd = isUserFd ? path : fs.openSync(path, options.flag || 'r', 0o666);

+  tryStatSync(fd, isUserFd);


Now that TF+I are in, can we start inlining all of these try* functions?

At least in a recently run benchmark (v8 5.8) it still seemed like it's a tiny bit better to keep them around. I suggest to wait until 6.0 has landed and if it's good to inline them, I'd make a single PR for all of them.

Ack, since it's an patch level enhancement that should land in v8.x, although it will probably be relevant for 1 minor node8 release.

lpinca · 2017-07-04T12:41:48Z

lib/fs.js

@@ -520,7 +520,6 @@ function tryStatSync(fd, isUserFd) {
  } finally {


I guess it was written like this for performance. If not, isn't cleaner to use catch and get rid of the threw variable?

I am about to open a separate PR that deals with lots of those :D

Get ready to run benchmarks then :) as I remember that some of these changes (finally -> catch) were rejected in the past. I think it was on timers.

Out of my perspective the main issue is that the error has to be thrown again.

Hm, when running the benchmark it does not seem to be a significant change, so it would be more churn than anything else.

@BridgeAR please note that the performance profile of try/catch/finally changed quite a bit since the code was added most likely :)

Of course but I ran the benchmarks with changing these to try catch and it did not show changed numbers. Therefore I think it's mainly churn to change it.

refack · 2017-07-08T21:42:00Z

lib/fs.js

@@ -552,10 +551,11 @@ fs.readFileSync = function(path, options) {
  var isUserFd = isFd(path); // file descriptor ownership
  var fd = isUserFd ? path : fs.openSync(path, options.flag || 'r', 0o666);

+  tryStatSync(fd, isUserFd);


Ack, since it's an patch level enhancement that should land in v8.x, although it will probably be relevant for 1 minor node8 release.

refack · 2017-07-08T21:50:56Z

CI: https://ci.nodejs.org/job/node-test-pull-request/9045/

refack · 2017-07-08T21:51:50Z

lib/fs.js

@@ -552,10 +551,11 @@ fs.readFileSync = function(path, options) {
  var isUserFd = isFd(path); // file descriptor ownership
  var fd = isUserFd ? path : fs.openSync(path, options.flag || 'r', 0o666);

+  tryStatSync(fd, isUserFd);
  // Use stats array directly to avoid creating an fs.Stats instance just for
  // our internal use.
  var size;


does a trinary with assign to const perform better?

There is not really a measurable difference.

🤷‍♂️ never mind then

* tryStatSync should not return any value If the function threw, it would never reach that code path. * only use try catch if necessary lchmodSync does not need the second try catch in most cases. PR-URL: nodejs#14055 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Refael Ackermann <refack@gmail.com>

refack · 2017-07-09T17:43:21Z

Extra quick sanity: https://ci.nodejs.org/job/node-test-commit-linuxone/7168/

* tryStatSync should not return any value If the function threw, it would never reach that code path. * only use try catch if necessary lchmodSync does not need the second try catch in most cases. PR-URL: #14055 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Refael Ackermann <refack@gmail.com>

MylesBorins · 2017-08-16T06:27:14Z

Should this be backported to v6.x-staging? If yes please follow the guide and raise a backport PR, if no let me know or add the dont-land-on label.

MylesBorins · 2017-09-19T16:02:57Z

ping re: v6.x

BridgeAR · 2017-09-19T20:58:30Z

I will open a PR to backport this soon.

MylesBorins · 2017-11-14T20:17:28Z

ping @BridgeAR

BridgeAR · 2017-12-11T06:13:27Z

When looking at it again there is little benefit for backporting, so I just changed the label accordingly.

nodejs-github-bot added the fs Issues and PRs related to the fs subsystem / file system. label Jul 3, 2017

TimothyGu reviewed Jul 4, 2017

View reviewed changes

lpinca reviewed Jul 4, 2017

View reviewed changes

lpinca approved these changes Jul 4, 2017

View reviewed changes

benjamingr approved these changes Jul 8, 2017

View reviewed changes

refack approved these changes Jul 8, 2017

View reviewed changes

refack self-assigned this Jul 8, 2017

refack reviewed Jul 8, 2017

View reviewed changes

refack force-pushed the micro-optimize-fs branch from ac7bd7c to 1806952 Compare July 9, 2017 17:42

refack merged commit 1806952 into nodejs:master Jul 9, 2017

addaleax mentioned this pull request Jul 11, 2017

v8.2.0 proposal #13744

Merged

MylesBorins added the lts-watch-v6.x label Aug 16, 2017

BridgeAR added the backported-to-v8.x label Sep 14, 2017

BridgeAR added dont-land-on-v6.x and removed lts-watch-v6.x labels Dec 11, 2017

MylesBorins mentioned this pull request Feb 27, 2018

fs: cleanup fd lchown and lchownSync #18329

Closed

2 tasks

refack removed their assignment Oct 12, 2018

BridgeAR deleted the micro-optimize-fs branch April 1, 2019 23:36

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fs: two minor optimizations #14055

fs: two minor optimizations #14055

BridgeAR commented Jul 3, 2017

TimothyGu Jul 4, 2017

BridgeAR Jul 4, 2017

refack Jul 8, 2017

lpinca Jul 4, 2017

BridgeAR Jul 4, 2017

lpinca Jul 4, 2017

BridgeAR Jul 4, 2017

BridgeAR Jul 4, 2017

benjamingr Jul 6, 2017

BridgeAR Jul 8, 2017 •

edited

Loading

refack Jul 8, 2017

refack commented Jul 8, 2017

refack Jul 8, 2017

BridgeAR Jul 9, 2017

refack Jul 9, 2017

refack commented Jul 9, 2017

MylesBorins commented Aug 16, 2017

MylesBorins commented Sep 19, 2017

BridgeAR commented Sep 19, 2017

MylesBorins commented Nov 14, 2017

BridgeAR commented Dec 11, 2017

		@@ -520,7 +520,6 @@ function tryStatSync(fd, isUserFd) {
		} finally {

fs: two minor optimizations #14055

fs: two minor optimizations #14055

Conversation

BridgeAR commented Jul 3, 2017

Checklist

Affected core subsystem(s)

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

BridgeAR Jul 8, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

refack commented Jul 8, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

refack commented Jul 9, 2017

MylesBorins commented Aug 16, 2017

MylesBorins commented Sep 19, 2017

BridgeAR commented Sep 19, 2017

MylesBorins commented Nov 14, 2017

BridgeAR commented Dec 11, 2017

BridgeAR Jul 8, 2017 •

edited

Loading