A overly technical view on the Glamour System

Quote:

---

You could say that we go chronologically through the list of items, and sort them like that, but that ends up adding extra load on… well… load times.

---

Sorry but none of what you said is hard, unless making a list of tuples is now hard ?

[(2054, TRUE),
(2055, FALSE),
(2056, FALSE),
...]

A list of tuples of length 16266 is not something that is hard to send and receive. Retrieving the boolean from that list is at most O(n), and O(1) if you use the appropriate data type. This is not something that stresses modern hardware. Not even if you do it a million times.

The reason they don't want to do it is because they are too lazy to code up the UI, that's it, that's all there is to it.

Please go actually benchmark what you're saying, go do it, even with the worst languages that isn't going to take longer than a microsecond.

Of course, the fact you have to wait a billion years between depositing two items in a company chest without the game yelling at you that multiple people are using the company chest (which is semi-bypassable if you don't use the UI) probably tells us that the real issue doesn't lie with what you're saying at all, a lot of it is with how bad the UI - and how that UI interacts with the server - in this game is.

Mao thinkings that might be good idea find out how GW2 does glams and then see if doables in FFXIV.

Quote:

---

Originally Posted by Jeeqbit

Well, they could be floats/decimal numbers.

---

That's my bad. I should have specified I meant numbers for identifiers.

Quote:

---

Originally Posted by Jeeqbit

They use a number. The ASCII is just a lodestone thing. Problem solved.

---

Good to know. It was more of a passing thought when I looked at the Lodestone, but I should have looked into it.

Quote:

---

Originally Posted by Jeeqbit

Yes, bit tables are sometimes used for things like this. What makes you believe they aren't doing it as efficiently as they can?

---

I guess I just looked at the different systems in the game, and wondered why the glamour system works the way it does. If you take into account that ARR was made while half the team was maintaining 1.0, it makes sense that a N-to-N solution would be taken, which would be easier to implement, but come with the limitations of the current glamour dresser. This is also why I added that last section, talking about how the solution makes sense on paper, but maybe not in practice, and that it may require some important changes. But, more than anything, I just wanted to try and get the ball rolling on a conversation about potential solutions.

Quote:

---

Originally Posted by Sindele

I don't mean this to be discouraging or insulting, but have you spent any time looking into the resources that exist for data mining and understanding both how the client operates and the data exchanged between client and server? There's an absolutely ridiculous amount of work that has already been done with regards to reverse engineering the client, and you could dispel some misconceptions very quickly with it.

---

Honestly, I wasn't aware of that. If you don't mind, could you link those?

Quote:

---

Originally Posted by Sindele

That aside, from a design standpoint, flagging on a per-item basis is wholly undesirable for one big reason (and one immediately apparent if you're familiar with similar systems like WoW transmog!): model overlap. You don't need to save a colossal bitmask for 16000 items when thousands of them are just duplicate models, nor would you want to.

---

To be fair, the transmog system was the first thing I thought of, since I already maxed out my dresser multiple times, and had to split hairs on what I'd keep and discard.

Quote:

---

Originally Posted by Sindele

Hexadecimal is always going to be more compact than the same value in base 10. As far as I know, lodestone IDs are just short hashes used to obfuscate the 'true' item IDs. Universalis don't care - Universalis pulls the ID straight from the game's item table.

---

So Universalis is using the real ID, while the hex is a fake to hide the real IDs. Got it.

Quote:

---

Originally Posted by FenyxRising

Sorry but none of what you said is hard, unless making a list of tuples is now hard ?

[(2054, TRUE),
(2055, FALSE),
(2056, FALSE),
...]

A list of tuples of length 16266 is not something that is hard to send and receive. Retrieving the boolean from that list is at most O(n), and O(1) if you use the appropriate data type. This is not something that stresses modern hardware. Not even if you do it a million times.

---

I suppose for me, it's more about the time it would take to load the items, UI frame, images for the items, building the tooltips, and most of the front end stuff. Splitting it by tabs was what I was trying to get to. Similar to what they do with the marketplace.

Quote:

---

Originally Posted by FenyxRising

The reason they don't want to do it is because they are too lazy to code up the UI, that's it, that's all there is to it.

---

They've been giving us a few QoL changes though. But with the scope of the game, I can't imagine that's an easy task for any team. Also, since UI is the first thing we see, they're always the first ones to get flak for things that might be out of their control, due to lacking data from the backend or other constraints.

Quote:

---

Originally Posted by FenyxRising

Please go actually benchmark what you're saying, go do it, even with the worst languages that isn't going to take longer than a microsecond.

---

To be fair, this was more a post to get a conversation started. And, seeing the response here, I'm glad it did just that. As for the time it takes, I should have mentioned that it wasn't just server times that I was looking into.

Quote:

---

Originally Posted by FenyxRising

Of course, the fact you have to wait a billion years between depositing two items in a company chest without the game yelling at you that multiple people are using the company chest (which is semi-bypassable if you don't use the UI) probably tells us that the real issue doesn't lie with what you're saying at all, a lot of it is with how bad the UI - and how that UI interacts with the server - in this game is.

---

The double user interaction is perhaps the reason that we can't have a glamour dresser as a housing item (I think I remember YoshiP mentioning that subject), but the amount of items stored should be a separate issue.

Quote:

---

Originally Posted by Eraden

Mao thinkings that might be good idea find out how GW2 does glams and then see if doables in FFXIV.

---

GW2 has a nice system, and I prefer their UI. That's what I had in mind when I made this post.

Quote:

---

Originally Posted by DaringDan

I guess I just looked at the different systems in the game, and wondered why the glamour system works the way it does.

---

1. They use physical machines rather than cloud servers. That means it's not really easily scalable like a cloud server. They decided against cloud servers because they are slower, despite the advantage of them being able to scale.
2. Having to transfer inventories to different physical servers seems a real challenge for them, so for example you don't carry your chocobo saddlebag into a duty server or your retainers to other worlds.
3. They are most likely taking into account network or database transfers over a large number of players at a given time as well.

I don't fully understand their reasons, because as an earlier poster stated, modern hardware is extremely good and it was even extremely good when I was doing tasks like this a decade ago. But they deal with a lot of users and seem to have their reasons.

As far as UI goes, we have experienced evidence of a disconnect between the UI team and... anyone else that knows how to code, in the past. They said it's "impossible" to show on tooltips when you have obtained items already, but there was definitely a way to do it if you really wanted to, even if the client got the information from the lodestone instead of directly from the server. And shortly after, they did it, proving their own argument wrong.

Maybe it is the UI team not working closely enough with programmers, but this really dented my confidence in how the UI team was being managed.

Quote:

---

If you take into account that ARR was made while half the team was maintaining 1.0

---

The glamour dresser itself was added after a few expansions, I recall, so it wasn't something we had in ARR. But we did have glamour prisms for as long as I can remember (although there used to be different types).

Quote:

---

Originally Posted by DaringDan

Honestly, I wasn't aware of that. If you don't mind, could you link those?

---

Nobody is going to link that sort of thing on the official forums. It's against the rules.

Glamours weren't implemented until 2.2 - early, but not at launch. I still have some old graded goldsmithing glamour prisms on an alt somewhere...

And yeah, won't be linking anything here. It's not exactly under wraps, though - poke around online and you'll find what you need.

Glamours are atrocious, were worse, but still is a terrible implemented system.
Specially when compared to the wardrobe in GW2.

We celebrating 2 color code-lines for clothes next expansion. GW2 since lunch has had 4.
The fact you can change any piece at any time but combat, stop and dye, without any hassle is just amazing.

In FFXIV glamouring is SUCH a process.