Search: count is 0? Or Search:first item is empty?

georgecollier · 2023-09-21T12:02:24.440Z

ntabs:

My question is will the actions wait for the searches to finish and output a yes or no before triggering each custom event?

Custom events run in sequence, always, I believe. That means if you have Trigger Custom Event 1 followed by Trigger Custom Event 2, Custom Event 2 will always run after Custom Event 1 has been completed.

ntabs · 2023-09-21T12:04:50.435Z

Thanks! greatly appreciate the swift reply @georgecollier

adamhholmes · 2023-09-21T12:20:36.383Z

ntabs:

Hi @adamhholmes did you hear back from Bubble’s support team yet?

Nope… heard nothing about this since June, and the issue still exists, so I can’t see it changing any time soon.

cmarchan · 2023-09-21T19:03:46.611Z

That is my understanding as well. There was a thread on order of operations some place in the forum where this was confirmed by Bubble.

tylerboodman · 2023-09-21T19:24:47.020Z

Do you know if it waits for the custom event to complete, or does it wait for all the actions to run?

Like if the former is true it is safe to Create a thing as the last step inside the Custom Event 1, then in Custom Event 2 search for the thing.

georgecollier · 2023-09-21T19:31:10.210Z

tylerboodman:

Do you know if it waits for the custom event to complete, or does it wait for all the actions to run?

What’s the difference? Surely a custom event is only complete when all actions inside the custom event have completed. That’s what I’ve always thought (maybe I just haven’t had a use case that’s proved me wrong though)

cmarchan · 2023-09-21T20:55:10.509Z

Notwithstanding why one would want to search a recently created item which can be “grabbed” via “result of step X” … and if the created object is very light (or you can first create it with no data inside it) you should be ok by searching for it in a subsequent custom event.

To answer your question my understanding is that the custom event is done when the actions have run. Bubble does not wait as it continues with its parallel processing where it can find space to do so.

shu.teopengco · 2023-10-22T14:50:07.590Z

Hi there, do you have tips on how to search elements that uses “search: count is 0” using the App Search Tool?

shu.teopengco · 2023-10-23T17:13:18.867Z

Anyone? Tips for finding using app search tool?

tylerboodman · 2023-10-23T17:26:31.469Z

No it seems the app search tool can only find “Do a search for” but not a specific expression like you’re trying to do…

shu.teopengco · 2023-10-24T00:09:09.904Z

It seems so. Can’t find it anywhere.

@josh
It might be a good idea if bubble will be able to search for these queries that can help optimize workload.

brenton.strine · 2024-01-20T15:03:44.176Z

Did this ever get fixed?

This is being actively discussed in the forum and Twitter.

igorracca13 · 2024-02-18T20:27:03.234Z

That is quite interesting, thank you for the insight!

johnny · 2024-02-22T20:16:19.484Z

Ok, so I emailed support@bubble.io, and this is what I understand:

search:first item is empty is more performant but generally uses more WUs compared to search:count < 1.

search:count < 1 general uses less WUs

The Bubble Support response is:

Search: first item is empty is usually the fastest approach. However, Search: count < 1 will usually use less workload.

And this is also intended behavior. Not sure why, but it is.

adamhholmes · 2024-02-22T23:30:15.356Z

More performant in what way?

johnny · 2024-02-22T23:58:15.115Z

I guess it’s just faster

adamhholmes · 2024-02-23T00:07:29.793Z

For what it’s worth… I just ran a simple test…

on a Database of over 100k items, searching for a single Thing by a text field value.

I ran a simple test… 10 times using search:first item is empty, and 10 times using search:count < 1 (using the exact same search for both)

Across 10 runs of each the average time to return the yes/no value was:

search:first item is empty = 338.9ms
search:count < 1 = 284.2ms

So search: count < 1 is faster than search: first item is empty, at least in my own tests (which is exactly what I would expect, as discussed already in this thread).

search: first item is empty is almost 20% slower, and costs more than 50% more in WU.

note: that’s in the case where there was an item to find.

I haven’t run the test when there isn’t one yet… I might try that next.

UPDATE:

So I ran the same test again, but this time with no item being found by the search.

search:count < 1 was again consistently faster than search: first item is empty, and this time by an even bigger margin.

Search: first item is empty = 382.9ms
Search: count < 1 = 279.7ms

(search: first item is empty was this time 38% slower than search: count < 1)

johnny · 2024-02-23T00:49:40.206Z

Hmm… yeah… i dunno thats just what bubble support told me

johnny · 2024-02-23T18:32:08.398Z

Update: got a follow up from support

I’ve heard from our engineering team that generally a search for a single item not being empty (Search:first item is/isn’t empty) is faster than aggregations - that’s because if the search is expensive (a lot of fields, a lot of data within those fields, etc.), then we stop early (as soon as we have 1 item) whereas if you do the whole count, it will scan through the entire search, possibly taking much longer.

That said, this is dependent on data that’s being searched. Since applications can be so different, I’d recommend testing out what seems to be most performant on average for your application and striking a balance between WUs and speed for your particular app.

devin.fraze · 2024-04-25T02:57:53.060Z

Hey @nick.carroll any update on this?