Whats going on with SE ?

[Ruf]

How would that work? The server has to find the drink you want, which means searching the data for it. At that point you're loading the data.

Theres a difference in between searching for a single item for a request versus asking for a whole 140 (including empty spaces)

I understand that. Why do you care? If in 2023 we have 140 inventory slots and four different types of saddlebags each with 140 more (a la FFXI), is that so much worse than having 700 inventory slots? Hell, it might be better because it'd be easier to organize the important stuff in the first four tabs instead of going through five times as much.

No one is denying that saddlebags are a workaround for this server limitation. What I'm asking is simple: so what?

I care, I'd rather have 700 inventory throught one window than a saddlebag & an app.

I mean, it's better than nothing. But it's not better than what could have been done without the server limitations being so severe. Of course, if they didn't make each raid tier have six different currencies that aren't counted as currencies, that might help too.

Ok so speak to that npc to get base armor (third option) then pick fourth option to get that item x20, then back out of menu go trade theses 20items for that single token with that other npc to the left & then come back to the first npc & then exchange it all for augment!, yea i'd rather deal without a labyrinth for a 10iLv. upgrade
Ps:Oh my bad sorry, more than one token is required & you are limited to one a week! (sure none of this route is accurate... but its pretty darn close) nxt patch rince & repeat

[Talraen]

Theres a difference in between searching for a single item for a request versus asking for a whole 140 (including empty spaces)

You're right, there is a difference: searching each time is far less efficient in the long run.

The server doesn't know where your item is, so it has to search blindly. In a bag with 140 items, assuming you have one stack of the item, it's going to have to load an average of 70 items to find the one you need. (And note, sorting is on the client side, so you can't assume the data is orderly at all on the server side.) If you use one item, that's better on average. If you use a second item, you're breaking even on average. Use more than that and you're behind. Even loading all 140 slots the first time you use any item is a much more efficient method. Loading them all up front is far better than either method for a variety of reasons.

With how clunky it is to organize things, not really. Pulling stuff off multiple retainers one at a time so I can craft something is not better than simply crafting it. In this day and age, a unified storage system and appropriate filters/searching would be better than having six different inventories (because housing storage is also a thing).

Added complexity isn't a good thing. It isn't accessible when doing things like crafting.

I mean, it's better than nothing. But it's not better than what could have been done without the server limitations being so severe. Of course, if they didn't make each raid tier have six different currencies that aren't counted as currencies, that might help too.

This is all true, but it's been true since 2.0 launched (to say nothing of 1.0's intensely less convenient inventory system). They aren't "adding complexity," and nothing is getting worse. In fact, between the extra 40 inventory slots and the stack size increase, you can hold many more items for crafting than you used to be able to.

So yes, it could be better, but the idea that it's going to do harm to the game when it's exactly the same as it's always been is a reach.

[Tridus]

The server doesn't know where your item is, so it has to search blindly. In a bag with 140 items, assuming you have one stack of the item, it's going to have to load an average of 70 items to find the one you need. (And note, sorting is on the client side, so you can't assume the data is orderly at all on the server side.) If you use one item, that's better on average. If you use a second item, you're breaking even on average. Use more than that and you're behind. Even loading all 140 slots the first time you use any item is a much more efficient method. Loading them all up front is far better than either method for a variety of reasons.

Indexes and fast search algorithms exist for just that. If they're simply doing a linear scan on your entire inventory when you search for one item specifically, then there's really no hope for the technical end of things. That said, getting your whole inventory makes the most sense because it can be done once, and searches after that can be done client side, where it's going to be blazing fast. (That's how the search feature in the game right now works.)

There's lots of reasons why loading the entire inventory client side is a good idea. It just doesn't need to be resynced constantly.

This is all true, but it's been true since 2.0 launched (to say nothing of 1.0's intensely less convenient inventory system). They aren't "adding complexity," and nothing is getting worse. In fact, between the extra 40 inventory slots and the stack size increase, you can hold many more items for crafting than you used to be able to.

So yes, it could be better, but the idea that it's going to do harm to the game when it's exactly the same as it's always been is a reach.

Leaving it the same when the rest of the market is advancing is effectively falling farther behind. The landscape today isn't what it was when 2.0 launched, and the less said about 1.0 the better. Improvements that reduce server load would open up a lot of options for them for advancements going forward beyond just better inventory systems, although those would be nice too.

Saying "it was good enough in 2014 so it's good enough in 2018" doesn't really hold water in the gaming industry.

[Remedi]

Saying "it was good enough in 2014 so it's good enough in 2018" doesn't really hold water in the gaming industry.

True, 1.0 was effectively based on that premise and we know how well that went

[Ruf]

You're right, there is a difference: searching each time is far less efficient in the long run.

The server doesn't know where your item is, so it has to search blindly. In a bag with 140 items, assuming you have one stack of the item, it's going to have to load an average of 70 items to find the one you need.

Oh please, its 2018, i dont believe in that statement at all whatsoever, with the spaghetti code maybe, but with a proper code, it should be much faster to pinpoint one item right away as long as the route to the search is flawless. (I in no way bite to that argument) well coded the search shouldnt even happen.

Indexes and fast search algorithms exist for just that.

saying that a proper code isnt possible is a lame excuse & its a pretty poor excuse for this game obviously with the programming power & catering this game have recieved already years ago, & now what do we get? crumbs for our money

[Tridus]

Oh please, its 2018, i dont believe in that statement at all whatsoever, with the spaghetti code maybe, but with a proper code, it should be much faster to pinpoint one item right away as long as the route to the search is flawless. (I in no way bite to that argument) well coded the search shouldnt even happen.

Except that it's actually not. If you have the whole inventory client side, you can search it in memory, locally, with no network IO at all. Nothing will be faster than that.

If you only actually needed one item client side, a search would be faster than retrieving the whole inventory, yes. But inventory is used so often that it doesn't make sense to do that 80 times when you could sync the whole thing on login and then do things locally (and send update events to the server). When something is needed so frequently, it becomes more efficient to eager load it and cache it than it does to load it on demand every time you need it.

For fishing, ruf is right in that GP is useless for gig fishing. But for normal fishing, gp is worthless beyond a certain point. You only need 650ish or so for that, or even less if you just straight up ignore fish that you know are worthless to you. GP even is a bit worthless for miner and botanist, 600 GP is all you need for collectables, and you can't really raise it enough to where it would help past that.

Higher tier nodes often penalize you if you only have 600 GP, like the 70 one star nodes IIRC need at least 650 to have any HQ chance whatsoever.

[Ruf]

Except that it's actually not. If you have the whole inventory client side, you can search it in memory, locally, with no network IO at all. Nothing will be faster than that.

If you only actually needed one item client side, a search would be faster than retrieving the whole inventory, yes. But inventory is used so often that it doesn't make sense to do that 80 times when you could sync the whole thing on login and then do things locally (and send update events to the server). When something is needed so frequently, it becomes more efficient to eager load it and cache it than it does to load it on demand every time you need it.




Higher tier nodes often penalize you if you only have 600 GP, like the 70 one star nodes IIRC need at least 650 to have any HQ chance whatsoever.

Have i said to not load the inventory completly at all? but everytime? no, it should be loaded one time completly then held, + to be honest, im not even sure,,, do you know how many items are useless in your inventory into the dungeons? In other terms, a slice&dice image could be created to cut half of the issue already imo & if that can be done, then yes, the search is minimized & so i really dont see why it cant be pinpointed, cache is a good point

[Talraen]

Oh please, its 2018, i dont believe in that statement at all whatsoever, with the spaghetti code maybe, but with a proper code, it should be much faster to pinpoint one item right away as long as the route to the search is flawless. (I in no way bite to that argument) well coded the search shouldnt even happen.

This isn't about how well-coded the search is, it's simple math and very simple theory. The server knows absolutely nothing about your inventory, because it hasn't loaded it. The inventory is not sorted or indexed, so you can't take advantage of any kind of tree searching or any other fancy methods. I would love to see your pseudocode for how you're finding something in a set you know absolutely nothing about faster than looking at each item in turn.

There's lots of reasons why loading the entire inventory client side is a good idea. It just doesn't need to be resynced constantly.

I couldn't agree more. I am baffled by why the data needs to be sent so often. I've only been responding to suggestions which I am qualified to point out the specific flaws in. As a programmer myself, armchair programmers make my life more difficult so I am biased against them.

[Ruf]

This isn't about how well-coded the search is, it's simple math and very simple theory. The server knows absolutely nothing about your inventory, because it hasn't loaded it. The inventory is not sorted or indexed, so you can't take advantage of any kind of tree searching or any other fancy methods. I would love to see your pseudocode for how you're finding something in a set you know absolutely nothing about faster than looking at each item in turn.



I couldn't agree more. I am baffled by why the data needs to be sent so often. I've only been responding to suggestions which I am qualified to point out the specific flaws in. As a programmer myself, armchair programmers make my life more difficult so I am biased against them.

Did you miss the part where i said core need to be fixed? If you say server this blablabla, then the server FAIL, simple as that, you keep reaching for excuses where as tridus & others said, it could be indexed, datas could be stored on client side to some extent so on so on, no matter how many excuses you come up with, how theres data centers issues & so on so on, the change is not impossible, if we went by impossible then nothing would be moving into society.

P.s: As far as the armchair programmers comment, are you programming while standing up?

[Talraen]

Did you miss the part where i said core need to be fixed? If you say server this blablabla, then the server FAIL, simple as that, you keep reaching for excuses where as tridus & others said, it could be indexed, datas could be stored on client side to some extent so on so on, no matter how many excuses you come up with, how theres data centers issues & so on so on, the change is not impossible, if we went by impossible then nothing would be moving into society.

Ah so the reality of how data works is an "excuse," I see. Uh, no. What you are actually proposing is literally impossible, because you don't appear to understand what you're proposing.

The specifics of the claims I was responding to notwithstanding, yes, there are ways they could make things more efficient. (Well, more than likely - I don't actually know enough about the system to be certain of that, but in as much as no system is perfectly optimized it's a very safe assumption.) They could, as you suggest, use your local storage to figure out where on the server the item you're accessing is most likely to be. That's not a bad idea for the reasons I laid out in my post; it's a bad idea for entirely different reasons which I addressed earlier in the thread. To sum up: it adds a lot of complexity for a small potential gain and a large potential loss. Loading everything up front is just a better idea for many reasons that myself and others have laid out in this thread. As Tridus and others have pointed out, the inefficiency here is the constant syncing, which you're not addressing at all.