benchmark: Fix hack by removing redundant + #17803

sreepurnajasti · 2017-12-21T05:41:21Z

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
commit message follows commit guidelines

Affected core subsystem(s)

Trott · 2017-12-21T06:35:48Z

Might it be best to remove the comment but leave the + in the two lines of code? I don't think the + is the "hack" that the comment is referring to. It's in pretty much all our benchmarks and is fairly idiomatic JavaScript for "make this thing a number". While the code probably works fine without the +, leaving it out will make this different than our other benchmarks while leaving it in makes it clear to someone reading the code what the expected type of those two configuration settings is.

BridgeAR · 2017-12-21T06:55:33Z

The "hack" was about actually having different internal n than the ones shown while running the benchmark.

Trott · 2017-12-21T07:02:00Z

The "hack" was about actually having different internal n than the ones shown while running the benchmark.

That makes a lot more sense. So perhaps the PR can be updated to try to fix that problem (which would be awesome 🎉) or updated to add more information to the comment so it's clear what it refers to (which would still be 👍).

AndreasMadsen · 2017-12-21T08:10:04Z

@Trott The value n in common.createBenchmark is a number, thus it is also a number in main. If you write typeof conf.n it will show "number".

At some point, 1.5 years ago, this was not the case and the type would be a string. This is no longer the case, thus we can safely remove the +.

AndreasMadsen · 2017-12-21T08:11:15Z

@Trott this is the code that makes it a number for all benchmarks: https://github.com/nodejs/node/blob/master/benchmark/common.js#L56

sreepurnajasti · 2017-12-21T08:42:39Z

@BridgeAR Seems like the testcase running correctly without any use of internal n. Correct me if I am wrong.

AndreasMadsen · 2017-12-21T09:04:50Z

@sreepurnajasti The benchmark tests are not very good, as they set all the values explicitly. Unfortunately, we can't afford to run the full benchmark in our tests.

AndreasMadsen

LGTM

Don't worry about the discusions. There is just some misunderstanding among us on how the benchmarks works, because they used to work differently.

BridgeAR · 2017-12-21T12:13:14Z

This comment (that I wrote, as I created this benchmark) was about n = 1e6 and not about anything else.

When you run the benchmark (e.g. node benchmark/run.js --filter deepequal-object.js assert) it will always show n set to this number, even though that is not true, since it will be reduced in the main part to have a similar running time for all size entries.

@AndreasMadsen it is clear that the number coercen is obsolete and that could be removed (when I wrote this benchmark I just copied the basic parts from other benchmarks and that was a common pattern, by now I already complain about this in newer benchmarks if someone does something similar) but this PR was about trying to fix the comment.

AndreasMadsen · 2017-12-21T12:34:52Z

@BridgeAR Ah, I see.

My comment in nodejs/code-and-learn#72 was about the number coercen. As you said it is obsolete but it is still in a lot of benchmarks that people copy, thus it keeps showing up. That is why I thought it would be excellent for a code-and-learn.

BridgeAR

Just to make sure this does not land as is.

Misunderstod purpose of TODO

Trott · 2017-12-21T20:34:01Z

@AndreasMadsen @BridgeAR So it sounds like the TODO should stay but the + can be removed here and elsewhere? (If so, that's basically, the exact opposite of what I originally suggested, so sorry about that.)

AndreasMadsen · 2017-12-21T22:49:55Z

@Trott completely agree.

Trott · 2017-12-21T23:14:16Z

(Oh, and maybe also: The TODO should be expanded to make it more clear what it is about.)

Trott

LGTM

Trott · 2017-12-21T23:16:53Z

@BridgeAR Do you have some suggested text if the TODO is restored? TODO: n should correspond to conf.n and not conf.n / size? Maybe followed by one sentence explaining why the division is currently necessary?

sreepurnajasti · 2017-12-22T07:11:25Z

@BridgeAR As per the discussion, I understood the following. Please, correct me if I misunderstood anything.

Number coercion i.e removal of + from the code, which is done in the present pr.
Todo:
In the code, we have internal n (conf.n) , external n (conf.n/size). Is the fix here is to configure the internal n at runtime? If so, can it be by defining theconst n = 1e6; conf.n = n / size;in main functionality with few other appropriate changes?

sreepurnajasti · 2018-01-04T07:24:01Z

@BridgeAR ping!!!

AndreasMadsen · 2018-01-04T07:54:04Z

This is what I wanted:

-  const size = +conf.size;
+  const size = conf.size;
// TODO: Fix this "hack"
- const n = (+conf.n) / size;
+ const n = conf.n / size;

sreepurnajasti · 2018-01-04T09:14:48Z

@AndreasMadsen Agreed. But, still the hack remains as is!! So can I even fix the hack in the same pr? Or should I make another for fixing the hack whereas this includes the above proposed changes?

AndreasMadsen · 2018-01-04T09:55:09Z

Agreed. But, still the hack remains as is!! So can I even fix the hack in the same pr? Or should I make another for fixing the hack whereas this includes the above proposed changes?

Both options are acceptable. Do what you are most comfortable with.

BridgeAR · 2018-01-04T09:58:26Z

@sreepurnajasti the code change should look like the one @AndreasMadsen suggested.

I am not aware of a good fix for my comment at the moment, otherwise I would have done that already.
size and n would have to be set as a single option together. We could do something like a string that looks like "n/size".

sreepurnajasti · 2018-01-04T10:35:24Z

@AndreasMadsen @BridgeAR Thanks. Done changes as per my understanding. Please review.

BridgeAR · 2018-01-04T11:35:00Z

@sreepurnajasti this does not fix the comment.

Right now, the best patch as far as I see it would be the one from @AndreasMadsen

-  const size = +conf.size;
+  const size = conf.size;
// TODO: Fix this "hack"
- const n = (+conf.n) / size;
+ const n = conf.n / size;

Fixes: nodejs/code-and-learn#72

sreepurnajasti · 2018-01-05T04:27:37Z

@BridgeAR Thanks. Its Done. Please review!!!

sreepurnajasti · 2018-01-05T04:35:21Z

benchmark/assert/deepequal-object.js

@@ -26,9 +26,9 @@ function createObj(source, add = '') {
 }

 function main(conf) {
-  const size = +conf.size;
+  const size = conf.size;


@BridgeAR Can I remove this line and replace size with conf.size as it is only used in two places(31,34)?

If you want to, yes. But I think it is not necessary.

sreepurnajasti · 2018-01-10T04:56:24Z

I just want to make sure that no other changes are needed. If so, can someone run CI please.

gireeshpunathil · 2018-01-10T05:03:50Z

CI: https://ci.nodejs.org/job/node-test-pull-request/12475/

BridgeAR · 2018-01-19T10:18:15Z

benchmark/assert/deepequal-object.js

@@ -26,9 +26,9 @@ function createObj(source, add = '') {
 }

 function main(conf) {
-  const size = +conf.size;
+  const size = conf.size;


If you want to, yes. But I think it is not necessary.

BridgeAR · 2018-01-19T10:24:56Z

Landed in f6ae451

PR-URL: nodejs#17803 Refs: nodejs/code-and-learn#72 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Ruben Bridgewater <[email protected]> Reviewed-By: Rich Trott <[email protected]>

BridgeAR · 2018-01-19T10:25:14Z

@sreepurnajasti thanks for being patient! :-)

sreepurnajasti · 2018-01-19T10:28:14Z

@BridgeAR @AndreasMadsen @Trott Thank you very much for assisting!!

AndreasMadsen · 2018-01-19T10:46:59Z

@sreepurnajasti Sorry, there was so much confusion. I'm glad it worked out :)

PR-URL: #17803 Refs: nodejs/code-and-learn#72 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Ruben Bridgewater <[email protected]> Reviewed-By: Rich Trott <[email protected]>

PR-URL: nodejs#17803 Refs: nodejs/code-and-learn#72 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Ruben Bridgewater <[email protected]> Reviewed-By: Rich Trott <[email protected]>

nodejs-github-bot added assert Issues and PRs related to the assert subsystem. benchmark Issues and PRs related to the benchmark subsystem. labels Dec 21, 2017

AndreasMadsen previously approved these changes Dec 21, 2017

View reviewed changes

cjihrig approved these changes Dec 21, 2017

View reviewed changes

BridgeAR requested changes Dec 21, 2017

View reviewed changes

Trott approved these changes Dec 21, 2017

View reviewed changes

sreepurnajasti force-pushed the deepequal-object branch from 7c5467a to c56b547 Compare January 4, 2018 10:31

benchmark: remove redundant +

994b9eb

Fixes: nodejs/code-and-learn#72

sreepurnajasti force-pushed the deepequal-object branch from c56b547 to 994b9eb Compare January 5, 2018 04:26

sreepurnajasti commented Jan 5, 2018

View reviewed changes

BridgeAR approved these changes Jan 19, 2018

View reviewed changes

BridgeAR closed this Jan 19, 2018

sreepurnajasti deleted the deepequal-object branch January 19, 2018 10:28

evanlucas mentioned this pull request Jan 30, 2018

v9.5.0 release proposal #18464

Merged

MylesBorins added dont-land-on-v6.x labels Feb 27, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

benchmark: Fix hack by removing redundant + #17803

benchmark: Fix hack by removing redundant + #17803

sreepurnajasti commented Dec 21, 2017

Trott commented Dec 21, 2017

BridgeAR commented Dec 21, 2017

Trott commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

sreepurnajasti commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

AndreasMadsen left a comment

BridgeAR commented Dec 21, 2017 •

edited

Loading

AndreasMadsen commented Dec 21, 2017 •

edited

Loading

BridgeAR left a comment

Trott commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

Trott commented Dec 21, 2017

Trott left a comment

Trott commented Dec 21, 2017

sreepurnajasti commented Dec 22, 2017

sreepurnajasti commented Jan 4, 2018

AndreasMadsen commented Jan 4, 2018

sreepurnajasti commented Jan 4, 2018 •

edited

Loading

AndreasMadsen commented Jan 4, 2018

BridgeAR commented Jan 4, 2018

sreepurnajasti commented Jan 4, 2018

BridgeAR commented Jan 4, 2018

sreepurnajasti commented Jan 5, 2018

sreepurnajasti Jan 5, 2018

BridgeAR Jan 19, 2018

sreepurnajasti commented Jan 10, 2018

gireeshpunathil commented Jan 10, 2018

BridgeAR Jan 19, 2018

BridgeAR commented Jan 19, 2018

BridgeAR commented Jan 19, 2018

sreepurnajasti commented Jan 19, 2018

AndreasMadsen commented Jan 19, 2018

benchmark: Fix hack by removing redundant + #17803

benchmark: Fix hack by removing redundant + #17803

Conversation

sreepurnajasti commented Dec 21, 2017

Checklist

Affected core subsystem(s)

Trott commented Dec 21, 2017

BridgeAR commented Dec 21, 2017

Trott commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

sreepurnajasti commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

AndreasMadsen left a comment

Choose a reason for hiding this comment

BridgeAR commented Dec 21, 2017 • edited Loading

AndreasMadsen commented Dec 21, 2017 • edited Loading

BridgeAR left a comment

Choose a reason for hiding this comment

Trott commented Dec 21, 2017

AndreasMadsen commented Dec 21, 2017

Trott commented Dec 21, 2017

Trott left a comment

Choose a reason for hiding this comment

Trott commented Dec 21, 2017

sreepurnajasti commented Dec 22, 2017

sreepurnajasti commented Jan 4, 2018

AndreasMadsen commented Jan 4, 2018

sreepurnajasti commented Jan 4, 2018 • edited Loading

AndreasMadsen commented Jan 4, 2018

BridgeAR commented Jan 4, 2018

sreepurnajasti commented Jan 4, 2018

BridgeAR commented Jan 4, 2018

sreepurnajasti commented Jan 5, 2018

sreepurnajasti Jan 5, 2018

Choose a reason for hiding this comment

BridgeAR Jan 19, 2018

Choose a reason for hiding this comment

sreepurnajasti commented Jan 10, 2018

gireeshpunathil commented Jan 10, 2018

BridgeAR Jan 19, 2018

Choose a reason for hiding this comment

BridgeAR commented Jan 19, 2018

BridgeAR commented Jan 19, 2018

sreepurnajasti commented Jan 19, 2018

AndreasMadsen commented Jan 19, 2018

BridgeAR commented Dec 21, 2017 •

edited

Loading

AndreasMadsen commented Dec 21, 2017 •

edited

Loading

sreepurnajasti commented Jan 4, 2018 •

edited

Loading