that setInterval

[titoBouzout]

Hello, thanks!,

Im very curious whats going on with the setInterval here https://github.com/anywhichway/nano-memoize/blob/master/src/nano-memoize.js#L60

Is called non-stop, really?

[titoBouzout]

The more functions you memo the more is it called......

[anywhichway]

Hmmm ... an uncaught design flaw I think. Has is actually cause you an issue with real code?

…

On Thu, Feb 14, 2019, 14:36 Tito ***@***.***> wrote: Hello, thanks!, Im very curious whats going on with the setInterval here https://github.com/anywhichway/nano-memoize/blob/master/src/nano-memoize.js#L60 Is called non-stop, really? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEuqSyew669YIXIWpoJ4d-l5MgyswPIBks5vNbqxgaJpZM4a8S6H> .

[titoBouzout]

Im just trying to be constructive ok,

Is causing issues with my brain, and that's were the real code gets executed.

why this needs to be called every 0.001 seconds? like in 1000 times per second, seriously?

do we really need a setInterval(whatever, 1) ? can this be set in a less intrusive way?

From a technical point of view I can imagine:

• battery ?
• browsers limiting setInterval because we do dumb stuff like this ? xD
• I can't really grasp the idea of the 1sec/1000 times , Im sorry

Can we also set a "global cleanup" check? because as you may know, if we set 10 MEMO function, then we gonna do 10k per second! !!!!!!!

Im kinda sort of mad because I believe ( I dont even remember) came here because of a Medium article "ohhhh we the best at memoize look our benchmarks" and you benchmarking the most dumb things and not even paying attention at how the code runs.......... so yeah..... wth is this ?

[anywhichway]

Good points. Will adjust the code to compute when the next expiration needs to be. Probably done in the next 48 hours.

…

On Thu, Feb 14, 2019, 21:02 Tito ***@***.***> wrote: Im just trying to be constructive ok, Is causing issues with my brain, and that's were the real code gets executed. why this needs to be called every 0.001 seconds? like in 1000 times per second, seriously? do we really need a setInterval(whatever, 1) ? can this be set in a less intrusive way? From a technical point of view I can imagine: - battery ? - browsers limiting setInterval because we do dumb stuff like this ? xD - I can't really grasp the idea of the 1sec/1000 times , Im sorry Can we also set a "global cleanup" check? because as you may know, if we set 10 MEMO function, then we gonna do 10k per second! !!!!!!! Im kinda sort of mad because I believe ( I dont even remember) came here because of a Medium article "ohhhh we the best at memoize look our benchmarks" and you benchmarking the most dumb things and not even paying attention the how the code runs.......... so yeah..... wth is this ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEuqS0-7ECi7EgLLfskh4Z1jKlPpS9NYks5vNhUygaJpZM4a8S6H> .

[titoBouzout]

Thank you, I would like to apologize for the way I wrote, Im sorry.

[anywhichway]

No worries. Note, I only wrote this code because someone else claimed they had written the fastest possible memoize and it was obviously inefficient, so it is ironic you found this flaw in my gnarly code (because I was also trying to make it tiny). The benchmarks are based on what others were using to make their claims of speed. I retrospect simply setting a regular timeout would have worked. Granted, this will result in two function calls for every memorized function. But, if it is critical the memos expire in a timely manner there is not much choice. Expiring the memos during the regular call will slow down the regular call for large memo sets. Alternatively a timeout to be set when the memo is created if it is supposed to expire. What are your thoughts? Get Outlook for Android<https://aka.ms/ghei36>

…

________________________________ From: Tito <notifications@github.com> Sent: Friday, February 15, 2019 12:42:27 PM To: anywhichway/nano-memoize Cc: Simon Y. Blackwell; Comment Subject: Re: [anywhichway/nano-memoize] that setInterval (#4) Thank you, I would like to apologize for the way I wrote, Im sorry. — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#4 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AEuqS7EXEnV6tVrA0BLo-r4UazT10G23ks5vNvGDgaJpZM4a8S6H>.

[titoBouzout]

I can think of :

1. just setting a setTimeout(cleanup, maxAge) when the memoize function is setup and has a maxAge option set.

I understand this has the downside that some objects will be very short lived, as a value set 100ms ago will be deleted if a timeout to cleanup runs, even if the timeout is set as high as 10 seconds, ex: if the value is saved at the 9.900sec mark it will be just deleted once we reach 10sec

To fix this we can save Date.now() also with the value. Then the cleanup function compares current Date.now() with the Date.now() saved within the value.

I believe the total point of this library is to reuse values and avoid computing expensive stuff. The point of maxAge I believe is so we don't leak or hold data forever, I mean is not time critical.

So I think is worthy to use the Date.now() approach because objects will live longer and avoid recomputing stuff. Even if it makes it slightly slower when maxAge used it will still be faster because an expensive function will be reused.

2. I was also thinking of a global interval, but this approach has many downsides. Stores for memoized functions should be global which will require special cleanup to not leak functions that are no longer used, Cleanup function will have to iterate over all the values of all the memoized functions set. Nevermind this.

Thanks!

[anywhichway]

@titoBouzout Fixed. Thanks for the assistance!

[titoBouzout]

ok lol. Just for the record of anyone reading, this is not fixed, he just moved the setInterval to the bottom and set the time of the interval in the variable expireInterval which still values 1. So nothing changed besides a bunch of variables renames. a5d42ae#diff-168726dbe96b3ce427e7fedce31bb0bcR44

[anywhichway]

Tito, the setInterval now only gets set on the redefined function not every time it gets called. Unless I am reading my own code incorrectly ... which is entirely possible. I am being pretty responsive here. Is there a reason you are being so negative? Regardless, I will take another look. It is kind of hard to write unit tests for this part of the code. I will put in some trace statements. I will reopen the issue when I get to a computer.

…

On Sat, Feb 16, 2019, 06:54 Tito ***@***.***> wrote: ok lol. Just for the record of anyone reading, this is not fixed, he just moved the setInterval to the bottom and set the time of the interval in the variable expireInterval which still values 1. So nothing changed besides a bunch of variables renames. a5d42ae #diff-168726dbe96b3ce427e7fedce31bb0bcR44 <a5d42ae#diff-168726dbe96b3ce427e7fedce31bb0bcR44> — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEuqS9DHyvvrVnjnwJojgcsRxDNu2lrZks5vN_FngaJpZM4a8S6H> .

[anywhichway]

@titoBouzout I have verified the fix is correct. If you wish to verify yourself, insert a trace statement below line 94 of src/nano-memoize.js and re-run the test suite. You will see the trace statement 4 times, but the functions are called far more frequently.

if(expireInterval) { console.log("Creating interval for " + fn); f.interval = setInterval(() => { // process key chngs out of cycle for speed for(const p in c) { if(maxAge) { tmout(c[p]); } delete c[p]; } },expireInterval); }