Avoid unnecessary selector evaluations #1273

josepot · 2019-05-02T15:43:31Z

Hi,

While I was working on #1272 I noticed that the selector was being evaluated more times than what's necessary. This PR makes sure that the selector is only evaluated when it's strictly needed.

cc: @MrWolfZ I would really appreciate you having a look at this, just to make sure that I didn't miss anything. All the previous tests are passing, and I have also added a new one to make sure that updating the selector works correctly. Please let me know what you think. Thanks!

Edit

A couple of things that I forgot to mention that perhaps aren't that obvious:

The only commit that's relevant for this PR is this one. The other one is the commit of Remove deps of useSelector #1272... I'm just being a bit optimistic and hoping that Remove deps of useSelector #1272 will get merged before this one.
My rationale for this change is the following:

If the update comes as a result of the forceRender of checkForUpdates, then there is no need to re-evaluate the selector.
If the update comes from "the outside", then it could be that:

The selector has been updated, in which case we need to re-evaluate the selector
It's a parent update that has nothing to do with us, so we don't need to re-evaluate the selector... What's likely, though, is that if the parent update was a result of the state changing, then checkForUpdates will already pick that change and trigger a render... Chances are that the child-update will have happened before the parent one, but it doesn't really matter, because it will happen during the same loop, so it will be batched... 🙂

MrWolfZ

This PR makes no sense in my opinion (and it bleeds changes from #1272), see my inline comment

src/hooks/useSelector.js

test/hooks/useSelector.spec.js

josepot · 2019-05-02T17:19:31Z

I will close this for now, sorry 😅

josepot · 2019-05-02T19:51:45Z

I'm reopening this PR.

I think that this should actually make a difference in terms of performance, because the selector will only be evaluated when it's strictly necessary. ~~Sure, it does add some complexity, but~~ the tests are all passing (now they are 🙈😅).

I guess that it would only make sense to merge this PR if the performance benchmarks are actually better, otherwise it wouldn't make sense to merge it, of course.

Please @MrWolfZ , could you please have another look at this 🙏? Thanks a lot!

MrWolfZ

I still think this change doesn't really add anything and makes the code just more complex. If the user cares about the selector logic being called too often, let them use a memoizing selector.

Also, since this change is supposed to bring better performance, please run this code against the benchmark repository.

src/hooks/useSelector.js

josepot · 2019-05-02T21:08:36Z

I still think this change doesn't really add anything and makes the code just more complex.

I honestly don't think that it makes the code much more complex than what it already was, but that is subjective, of course.

If the user cares about the selector logic being called too often, let them use a memoizing selector.

But this is the thing, even if the user memoized the selector, the selector would be executed twice when an update is detected and when the render that happens after that takes place... Memoized selectors are cheap, but they are not free, in the case of reselect the whole tree of selectors has be be re-evaluated with their latest arguments, if we can avoid that I think that we should. Not to mention the fact that lots of ppl don't memoize their selectors.

Also, without this possible improvement, any time that there is an update from an ancestor, if that update is not changing the selector, the selector would be evaluated again... Which is totally unnecessary. Honestly, I don't understand why you are so against this.

Also, since this change is supposed to bring better performance, please run this code against the benchmark repository.

I would love to do that. ~~Could you please show me where that repo is?~~ Found it! https://github.com/reduxjs/react-redux-benchmarks

josepot · 2019-05-02T21:29:34Z

Also @MrWolfZ, please look at the bright side. Even if this change is actually bad or unnecessary, at the very least it will help us to come up with more and better tests for possible future refactors. So far we already got one new test 😉

MrWolfZ · 2019-05-04T07:28:49Z

The latest version of this code looks much better, and I think it is now correct. However, I am still doubtful whether it's worth it. Here's why:

it breaks the "do the simplest thing that can work" principle
it does objectively increase the complexity of the code (and how could it not if it does more than the original code?); you can see by running both versions through jshint that the number of statements increases and that the cyclomatic complexity doubles. There are also two examples from the code I can give that correspond to this.
1. the current version has two if statements, while the proposed code has five
2. the current code can be read from top to bottom in a linear fashion, while in the proposed version for example the variable result is declared in line 57, potentially mutated multiple times in the try block starting at line 96, but accessed via closure in line 60 again, so it needs some mental jumps to follow the code
The reason I go on so much about complexity is that I want to prevent this code from becoming too much like connectAdvanced. Have you looked at it recently? Who even understands that anymore except a handful of people?
the proposed solution involves caching, and as we all know, there are only two hard things in computer science. While the code looks correct to me, are we really sure we have thought about all the possible edge cases?
it seems like a premature optimization that can easily be added later if we get reports about performance issues
while the synthetic benchmark shows there are situations in which it definitely performs better, in reality you would probably never use that pattern, and if you would, you would use a memoizing selector (@josepot mentioned there is still an associated cost with evaluating those, which is of course true, but I would like to see at which point it actually becomes an issue; I think you'll need selectors nested incredibly deeply for this to ever become noticeable)

But again, the code seems correct, and as demonstrated it can provide performance improvements. I can absolutely see both sides of the argument. Since I am not the maintainer of this repo, I can't decide this in any case. It is up to the maintainers to decide whether they prefer simpler more maintainable code or more performant code.

josepot · 2019-05-04T15:11:22Z

Edit

In here there was an answer to @MrWolfZ comment that now I wish I had handled better. That's why I'm making this final edit. If you are curious to see that comment, then use the "edit" dropdown. 🙂

timdorr · 2019-05-08T17:46:13Z

Kind of an obvious question: Why would the selector ever change? I'm pretty sure Dan's hand will emerge from your computer monitor and slap you in the face if you ever did this:

useEffect(user_id ? loadUserEffect : loadLoginEffect)

While not enforced by a eslint rule, that kind of thing is basically the same as putting your Hooks inside a conditional. It's objectively bad code and should feel bad.

I think that kind of makes this change a moot point, no?

josepot · 2019-05-08T19:36:24Z

@timdorr

Kind of an obvious question: Why would the selector ever change?

For instance:

function User({ id }) {
  const getUser = useCallback(state => state.users[id], [id]);
  const user = useSelector(getUser);
  // return some elements to show the user-data here
}

It's quite likely that the id will also be the key of the Component, in which case the id would never change, of course... But check this out:

function UsersDiff({ idA, idB }) {
  const getUserA = useCallback(state => state.users[idA], [idA]);
  const getUserB = useCallback(state => state.users[idB], [idB]);

  const userA = useSelector(getUserA);
  const userB = useSelector(getUserB);

  // return some elements that show the differences or whatever
}

In the latest case, the IDs could change and the key of the component wouldn't be based on those IDs.

The only reason why the selector could change is that this selector -unlike the selector of connect- only takes the state, and sometimes you would like to access some props...

I'm pretty Dan's hands will emerge from your computer monitor and slap you in the face if you ever did this

I'm having a hard time picturing Dan slapping someone in the face, but I guess that I get what you meant 😄

josepot · 2019-05-18T16:52:28Z

Hi @MrWolfZ @timdorr and @markerikson,

Sorry that it took me a while to post the results. I got started at a new job and I have been quite busy for the last 2 weeks.

These results are obtained with the code of this branch. You may want to have a look at it just to make sure that I didn't screw anything.

If I'm reading these results correctly, they are showing an improvement on performance and that is despite the fact that the selectors of the benchmarks don't actually perform any heavy computations. These are the results:

Running benchmark deeptree                                                                                                                                                            
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation                                           
    Checking max FPS... (30 seconds)              
    Running trace...    (30 seconds)                                                                 
  react-redux version: 7.0.0.alpha-useSelectorNoDeps                                                 
    Checking max FPS... (30 seconds)                                                                 
    Running trace...    (30 seconds)                                                                 
                                                                                                     
Results for benchmark deeptree:                                                                      
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬─────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                              │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                         │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼─────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 44.06   │ 86.9, 0.6    │ 9452.44   │ 10396.98  │ 6063.76  │ 56,57,58,55,57,58,57,58,41,35,37,34,35,34,33,24,27,24,24                │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼─────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 42.27   │ 88.8, 0.6    │ 9521.25   │ 10229.33  │ 6096.60  │ 56,55,56,55,56,53,54,53,56,52,53,37,33,37,33,34,33,32,24,21,23,22,25,25 │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴─────────────────────────────────────────────────────────────────────────┘
Running benchmark deeptree-nested                                                                    
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation                                     
    Checking max FPS... (30 seconds)                                                                 
    Running trace...    (30 seconds)             
  react-redux version: 7.0.0.alpha-useSelectorNoDeps
    Checking max FPS... (30 seconds)         
    Running trace...    (30 seconds)         
                                                  
Results for benchmark deeptree-nested:          
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬────────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                                 │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                            │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 56.75   │ 128.6, 0.3   │ 3937.45   │ 2915.23   │ 1803.65  │ 58,57,60,59,60,59,56,58,59,58,60,59,58,57,60,57,55,54,53,55,52,51,49,53,53 │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 56.87   │ 126.1, 0.4   │ 3993.32   │ 2951.71   │ 1872.79  │ 58,60,59,60,58,57,58,60,59,58,59,58,56,59,56,53,55,53,56,55,51,54,54       │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴────────────────────────────────────────────────────────────────────────────┘
Running benchmark forms                                                                              
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation
    Checking max FPS... (30 seconds)                                                                 
    Running trace...    (30 seconds)                                                                 
  react-redux version: 7.0.0.alpha-useSelectorNoDeps             
    Checking max FPS... (30 seconds)                                                                 
    Running trace...    (30 seconds)                                                                 
                                                  
Results for benchmark forms:         
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬─────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                              │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                         │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼─────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 56.26   │ 943.0, 0.2   │ 6138.50   │ 972.85    │ 2727.64  │ 55,57,56,55,57,55,57,56,57,60,56,57,56,53,58,56,57,55,57,55,56,55,55    │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼─────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 55.58   │ 898.3, 0.3   │ 6272.87   │ 1015.13   │ 2758.28  │ 56,57,56,55,56,55,57,56,57,56,57,56,57,52,58,57,53,56,54,56,55,50,55,55 │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴─────────────────────────────────────────────────────────────────────────┘
Running benchmark stockticker                                                                        
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)
  react-redux version: 7.0.0.alpha-useSelectorNoDeps
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)

Results for benchmark stockticker:
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬────────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                                 │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                            │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 38.37   │ 170.5, 3.3   │ 18877.63  │ 7037.41   │ 2102.51  │ 40,39,41,38,39,40,38,39,41,38,39,38,39,40,38,37,36,38,37,36,36             │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 37.07   │ 168.9, 3.3   │ 19129.98  │ 6931.06   │ 2035.21  │ 39,40,42,39,37,38,37,36,39,35,36,37,35,34,37,38,39,37,35,36,37,34,36,37,37 │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴────────────────────────────────────────────────────────────────────────────┘
Running benchmark tree-view
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)
  react-redux version: 7.0.0.alpha-useSelectorNoDeps
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)

Results for benchmark tree-view:
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬───────────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                                    │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                               │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼───────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 50.04   │ 426.4, 0.9   │ 5597.07   │ 12955.30  │ 463.26   │ 55,39,53,55,42,48,53,52,54,41,52,46,49,50,54,51,55,52,54,52,48,51,48,51,50,50 │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼───────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 48.20   │ 422.7, 1.1   │ 6539.58   │ 9667.43   │ 544.14   │ 42,49,53,44,49,51,55,52,43,50,47,44,46,49,52,47,53,44,49,43,55,50,51,44,53,53 │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴───────────────────────────────────────────────────────────────────────────────┘
Running benchmark twitter-lite
  react-redux version: 7.0.0.alpha-avoidSelectorEvaluation
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)
  react-redux version: 7.0.0.alpha-useSelectorNoDeps
    Checking max FPS... (30 seconds)
    Running trace...    (30 seconds)

Results for benchmark twitter-lite:
┌─────────────────────────────────────┬─────────┬──────────────┬───────────┬───────────┬──────────┬──────────────────────────────────────────────────────────────────────────────────┐
│ Version                             │ Avg FPS │ Render       │ Scripting │ Rendering │ Painting │ FPS Values                                                                       │
│                                     │         │ (Mount, Avg) │           │           │          │                                                                                  │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼──────────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-avoidSelectorEvaluation │ 47.54   │ 2.2, 0.5     │ 21001.58  │ 3599.72   │ 382.40   │ 59,60,59,60,58,59,60,58,56,50,47,44,40,39,34,29,28,27,27                         │
├─────────────────────────────────────┼─────────┼──────────────┼───────────┼───────────┼──────────┼──────────────────────────────────────────────────────────────────────────────────┤
│ 7.0.0.alpha-useSelectorNoDeps       │ 47.39   │ 2.1, 0.5     │ 21157.33  │ 3660.14   │ 403.49   │ 59,60,58,59,58,59,56,59,58,59,55,57,55,54,48,46,44,41,38,34,33,32,31,30,29,28,28 │
└─────────────────────────────────────┴─────────┴──────────────┴───────────┴───────────┴──────────┴──────────────────────────────────────────────────────────────────────────────────┘
Done in 862.88s.

I have also tried benchmarking the same, but making the selectors perform some silly work on purpose, and as I already thought the results in those cases show a much more significant improvement in performance in favor of the code of this PR.

However, I think that just with these results we should consider merging this PR 🙂

markerikson · 2019-05-18T19:14:55Z

FWIW, FPS values will fluctuate from run to run. Anything +-3-ish FPS should be basically considered equal.

josepot · 2019-05-18T22:19:41Z

FWIW, FPS values will fluctuate from run to run. Anything +-3-ish FPS should be basically considered equal.

@markerikson

Sure, this optimization is only meant to be significant for those selectors that perform computations. All the selectors of our benchmarks are trivial selectors that only select things directly from the state... What I meant by showing these results is that this change can only make things better, never worse.

Do you want me to share the results and the code that test how this change would perform on non-trivial selectors?

Do you want me to create a new benchmark that has some non-trivial selectors?

Do you want me to just close this PR given those results?

Thanks!

markerikson · 2019-05-18T22:35:37Z

I haven't made up my mind on this one overall. Still trying to get a better understanding of the tradeoffs and weighing options. So yes, more benchmarks, examples, and edge cases would be helpful.

markerikson · 2019-05-18T23:14:49Z

@josepot : hmm. apologies, I'm getting the PRs mixed up :)

Gimme some time to sit down and actually evaluate this one in the next few days.

markerikson · 2019-05-19T23:47:48Z

Awright. I don't see any harm in this. I did swap the ternary for an if/else, though. Not a ternary fan.

* Avoid unnecessary selector evaluations * Clean up state assignment logic * Add missing shallowEqual export

MrWolfZ suggested changes May 2, 2019

View reviewed changes

src/hooks/useSelector.js Outdated Show resolved Hide resolved

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch from 1de42aa to 5ee5de3 Compare May 2, 2019 17:10

lukaszfiszer reviewed May 2, 2019

View reviewed changes

test/hooks/useSelector.spec.js Outdated Show resolved Hide resolved

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch from 5ee5de3 to 32d81d5 Compare May 2, 2019 17:17

josepot closed this May 2, 2019

josepot reopened this May 2, 2019

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch from 32d81d5 to 71dfe44 Compare May 2, 2019 19:43

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch 2 times, most recently from c332fa3 to 5d276ae Compare May 2, 2019 20:20

MrWolfZ reviewed May 2, 2019

View reviewed changes

src/hooks/useSelector.js Outdated Show resolved Hide resolved

src/hooks/useSelector.js Outdated Show resolved Hide resolved

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch 7 times, most recently from 10cf5f4 to 88dce64 Compare May 3, 2019 09:07

josepot mentioned this pull request May 4, 2019

Feedback: React-Redux v7 hooks alpha #1252

Closed

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch 3 times, most recently from a208a7b to d129ef1 Compare May 4, 2019 13:49

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch 3 times, most recently from 95abfb4 to 28f928a Compare May 5, 2019 16:10

Avoid unnecessary selector evaluations

5ab306b

josepot force-pushed the feat/useSelector-avoid-unnecesary-selector-evaluations branch from 8d626fd to 5ab306b Compare May 18, 2019 16:10

josepot mentioned this pull request May 19, 2019

adjust the hooks docs with more details about useSelector and add some more examples #1267

Merged

markerikson added 2 commits May 19, 2019 19:44

Clean up state assignment logic

d381e26

Add missing shallowEqual export

4dc56d5

markerikson merged commit 007b5c7 into reduxjs:v7-hooks-alpha May 19, 2019

timdorr pushed a commit that referenced this pull request May 30, 2019

Avoid unnecessary selector evaluations (#1273)

3ef4ac5

* Avoid unnecessary selector evaluations * Clean up state assignment logic * Add missing shallowEqual export

timdorr pushed a commit that referenced this pull request Jun 11, 2019

Avoid unnecessary selector evaluations (#1273)

4d3a7e1

* Avoid unnecessary selector evaluations * Clean up state assignment logic * Add missing shallowEqual export

This was referenced Sep 3, 2019

[Snyk] Upgrade react-redux from 7.1.0 to 7.1.1 xmartinezpujol/frontend-quiz#19

Merged

[Snyk] Upgrade react-redux from 7.1.0 to 7.1.1 Niweera/wordhunter#8

Open

louiszawadzki mentioned this pull request Sep 9, 2019

useSelector returning out-of-date result #1316

Closed

This was referenced Oct 26, 2019

[Snyk] Upgrade react-redux from 5.0.7 to 7.1.1 ajesse11x/redux#3

Open

[Snyk] Upgrade react-redux from 5.0.6 to 7.1.1 ajesse11x/AspNetCore#3

Open

snyk-bot mentioned this pull request Nov 25, 2019

[Snyk] Upgrade react-redux from 5.0.7 to 7.1.3 ajesse11x/redux#9

Open

snyk-bot mentioned this pull request Dec 19, 2019

[Snyk] Upgrade react-redux from 5.0.7 to 7.1.3 Itchibon777/freeCodeCamp#11

Open

snyk-bot mentioned this pull request Feb 5, 2020

[Snyk] Upgrade react-redux from 7.0.1 to 7.1.3 angelika-draczynska/learning-rest-api#18

Open

snyk-bot mentioned this pull request Mar 9, 2020

[Snyk] Upgrade react-redux from 7.0.2 to 7.2.0 danactive/history#407

Closed

snyk-bot mentioned this pull request Mar 27, 2020

[Snyk] Upgrade react-redux from 7.0.1 to 7.2.0 angelika-draczynska/learning-rest-api#25

Open

This was referenced Apr 9, 2020

[Snyk] Upgrade react-redux from 7.0.1 to 7.2.0 civilizador/wzr-redux#1

Open

[Snyk] Upgrade react-redux from 7.0.3 to 7.2.0 civilizador/old-friends-pizza#8

Open

snyk-bot mentioned this pull request May 7, 2020

[Snyk] Upgrade react-redux from 7.0.1 to 7.2.0 blockchain/blockchain-wallet-v4-frontend#2355

Closed

albertodev7 pushed a commit to albertodev7/react-redux that referenced this pull request Dec 8, 2022

Avoid unnecessary selector evaluations (reduxjs#1273)

48f6875

* Avoid unnecessary selector evaluations * Clean up state assignment logic * Add missing shallowEqual export

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Avoid unnecessary selector evaluations #1273

Avoid unnecessary selector evaluations #1273

josepot commented May 2, 2019 •

edited

Loading

MrWolfZ left a comment

josepot commented May 2, 2019

josepot commented May 2, 2019 •

edited

Loading

MrWolfZ left a comment

josepot commented May 2, 2019 •

edited

Loading

josepot commented May 2, 2019

MrWolfZ commented May 4, 2019

josepot commented May 4, 2019 •

edited

Loading

timdorr commented May 8, 2019 •

edited

Loading

josepot commented May 8, 2019 •

edited

Loading

josepot commented May 18, 2019 •

edited

Loading

markerikson commented May 18, 2019

josepot commented May 18, 2019 •

edited

Loading

markerikson commented May 18, 2019

markerikson commented May 18, 2019

markerikson commented May 19, 2019

Avoid unnecessary selector evaluations #1273

Avoid unnecessary selector evaluations #1273

Conversation

josepot commented May 2, 2019 • edited Loading

Edit

MrWolfZ left a comment

Choose a reason for hiding this comment

josepot commented May 2, 2019

josepot commented May 2, 2019 • edited Loading

MrWolfZ left a comment

Choose a reason for hiding this comment

josepot commented May 2, 2019 • edited Loading

josepot commented May 2, 2019

MrWolfZ commented May 4, 2019

josepot commented May 4, 2019 • edited Loading

Edit

timdorr commented May 8, 2019 • edited Loading

josepot commented May 8, 2019 • edited Loading

josepot commented May 18, 2019 • edited Loading

markerikson commented May 18, 2019

josepot commented May 18, 2019 • edited Loading

markerikson commented May 18, 2019

markerikson commented May 18, 2019

markerikson commented May 19, 2019

josepot commented May 2, 2019 •

edited

Loading

josepot commented May 2, 2019 •

edited

Loading

josepot commented May 2, 2019 •

edited

Loading

josepot commented May 4, 2019 •

edited

Loading

timdorr commented May 8, 2019 •

edited

Loading

josepot commented May 8, 2019 •

edited

Loading

josepot commented May 18, 2019 •

edited

Loading

josepot commented May 18, 2019 •

edited

Loading