Improvement in implementation of “:count is 0”

Announcements
21

We shouldn’t have to ask for this kind of stuff to be priorities. THEY ARE. Everything related to WU improvement should have been done a lot more time ago. We started to see improvement probably around June with the SAWOL update. But a lot of requests was already created (and answered by @josh) when the WU pricing was released. This one about 0 issue was raised a lot of time ago and known from team. Returning only fields needed in search was requested a lot of time and never implemented even if @josh agree to this. We get some updates on logs and charts a week ago. And even so, we are far from what we could expect to manage WU and optimize them.

From my point of view, for 1 year, the top priorities should have been: stability, and WU improvement

And I hope they don’t consider this actually fixed…

9 Likes
22

I think ever Bubbler has their own list of what they consider to be the highest priority items, and I guarantee you that they’re all different. And internally they’re also working to advance Bubble with new features plus fix issues that we have no idea about at all.

It’s actually a monumental task to take in all the various competing needs and opinions and find the most impactful stuff to work on. In fact, if you’re up for this kind of work, it would actually be really helpful!

I’ve never been at any company that gets it 100% right, but I do feel that Bubble is better at it now then they were 1 year ago.

3 Likes
23

Have you done this with “first item is empty”? We use that one primarily. Curious if it’s as efficient as “count is [positive integer].”

1 Like
24

Do a search:first item = 0.3 WU because it’s a search that attempts to return Things

Count is {positive integer} = 0.2 WU as it’s an aggregate query

4 Likes
25

great job for listening to feedback again. keep it coming!

1 Like
26

It’s not… it’s ALWAYS at least 50% more costly in WU, and in some cases a lot more - in my original tests, discussed in one of the linked threads, first item is empty cost over 11 times more WU than count < 1 (but that will always be dependant on the data size).

(it doesn’t have to be a positive integer by the way… any number other than 0, positive or negative, integer or decimal… the issue is specific to the literal value of 0).

WU cost aside… there is an argument to be made that first item is empty (despite costing 50% more in WU) is actually faster than count < 1 (Bubble support have been fairly adamant about this when I’ve discussed it with them).

But in my own testing I’ve never observed that to be true. I’ve tested this with a few hundred thousand items - maybe with a few million it might be different - but in all of my own testing using the count not only costs less, but is faster as well.

4 Likes
27

That is really good to know! Thanks. I’ll put it in our notes to slowly replace all our “first item is empty” arguments, then.

This WU pricing is about to go kick in for us on 10/1, and our cost will almost triple (currently on legacy Pro plan plus one capacity unit). Saving WU anywhere we can helps.

2 Likes
28

I think actually @georgecollier offer support to help people about WU optimization.

2 Likes
29

Thanks @allenyang , does this mean that count is 0 is actually faster and less WU-intensive than count < 1? Or would the difference be negligible? I don’t want to update 50 different expressions if I don’t have to. Would really appreciate a response on this.

Edit:

Oh, there’s my answer.

2 Likes
30

My confusion is confused! :thinking:

1 Like
31

Here’s the summary:

  • count < 1 is still best
  • Now count = 0 is second-best
  • :first item is empty is now the worst
19 Likes
32

Well, what’s inside the typical bubble? Same thing in the bubbleheads. :wink:

1 Like