PLD/BLU

Eorzea Time
 
 
 
Language: JP EN FR DE
Version 3.1
New Items
users online
Forum » FFXI » Jobs » Paladin » PLD/BLU
PLD/BLU
First Page 2 3 4 5 6 7 8
 Asura.Cambion
Offline
Server: Asura
Game: FFXI
user: Cambion
Posts: 415
By Asura.Cambion 2021-05-25 22:11:13
Link | Quote | Reply
 
Bahamut.Balduran said: »
1: So firstly, let's agree that to date, we have no evidence from testing that SIRD can be interrupted in your scenarios 1 & 2, that is a fact.

2: At this point, it is fair to conclude that scenario 3 is not equivalent to 1 and 2, and not the intended behavior of GearSwap. Based on this summery, there is no evidence to tie the 3 scenarios together and say they should all work the same way, because they don't.

3: Also, Martel pointed out on the previous page that...

4: I would reiterate that FF11 is not designed for macro spamming in all respects, and to conclude, there yet remains any evidence that debunks precast > midcast affecting any sird, without producing an anomaly through deliberate spamming before the spell is cast.

1: I disagree with this. Multiple people have stated, and I'm sure hundreds more have experienced, an interrupt when they didn't think it should be possible. But it happens so rarely that we write it off as lag, player error, accidental movement, or maybe a mob spell/move that caused it. Unless it results in something catastrophic, not of us are scrolling up to investigate, we recast and move on with whatever we're tanking.

2: Gearswap is 0s and 1s. All it sees, knows, and processes are current gear > pre cast gear > mid cast gear > after cast gear. It doesn't matter what we're doing, or what the original status was, it's logical processing is the same for all 3 scenarios. There is zero evidence to support that gearswap handles gearswaps differently based on the original status the command was triggered from. Input and output.

3&4: You're discussing macro spamming and inefficiencies. You have to understand the goal posts have been moved twice thus far.
The discussion at hand was whether or not, a pre-cast set could ever effect the outcome or success of a spell. The current belief is people saying 'no' because 'it's handled instantly in the same packet'. My counter theory, was that at some point, even if only for a fraction of a second, your precast gear has to be equipped during the casting of the spell. The response I got was basically 'no' because pre cast and miscast were instant, same packet, simultaneous, etc etc.
So, first I do the test, and they blame after-cast sets. I eliminate that variable in the test, now they blame the packet cap. Both are valid arguments and justify re-testing, they are correct.

My test was to disprove that pre-cast and midcast are handled 'instantly' 'simultaneously' or 'same packet'. I proved there is a void between pre-cast and mid-cast in which a hit, check, and interrupt can occur. But, I likely created a flaw in the test, by capping the packet queue, so a new test needs to be done.

Now, the NEW theory that needs to be tested, is whether or not, that occurred because I over-capped the packets sent OR whether because there was a legitimate period of time, in which the game was calculating my cast, while in precast gear, and being hit.

I'll try that, and see what happens. If I still find an interrupt, we'll see what the next adjustment of the statement is. Hopefully an interrupt doesn't happen and we just move on. It really doesn't matter, regardless of the outcome, my lua is already built to maximize both scenarios with a simple keystroke. But my curiosity is peaked... piqued?

I've gotta knock out Omen and Odyssey first, but after that I'll give it a shot.
Offline
Posts: 1109
By DaneBlood 2021-05-25 22:52:01
Link | Quote | Reply
 
i tried typing this 3 time and close the tab before saving. it to late and im tire so ima make it short


Asura.Cambion

Tou are right we need to be carefull about moving the goal post.
however we also needto be crarefull to not jump to conclussions

3&4
You have not proven there is a time between precast and midtcast.
that is stil la thery

You have proven that there is an additional effect that affect SRID negatively besides aftercast.
What the effect is is still in theory.

it might as thorny was in just as well be that due to the button smashing you gear is not equiping correctly.
and in this case just applying the SIRD in precast might not be the solutions due to aftercast still being a thing.

However the time between precast to midtcast was/is just a stepping stone to the final argument. do you need sird in precast, for it to work ( while ignoring aftercast?).

This still needs to be proven


WE might not be able to fully prove whatever there is a time between precast and mditcast as this all depends finally on the server handling the data and we do not have access to it.

But we should be able to narrow down if SIRD is needet in precast if button smashing vs if not button smashing. and if so whatever it is needed in after cast as well



2. im not sure what is tryingto be said here and it might just be me beeing tired.
but button smashing and non buttong smashing should not be classifed together s we do have a plausible theory of a factor that changes the behavioer (the outgoung queue)
if the soruce of the problme you expirence is different in those scenarios we cannot draw conclussion from one to another.

not sure if i got the right gist here


Just my 50 cent
and good job on following up with a test
 Bahamut.Balduran
Offline
Server: Bahamut
Game: FFXI
user: balduran
Posts: 270
By Bahamut.Balduran 2021-05-26 05:20:40
Link | Quote | Reply
 
Asura.Cambion said: »
1: I disagree with this. Multiple people have stated, and I'm sure hundreds more have experienced, an interrupt when they didn't think it should be possible. But it happens so rarely that we write it off as lag, player error, accidental movement, or maybe a mob spell/move that caused it. Unless it results in something catastrophic, not of us are scrolling up to investigate, we recast and move on with whatever we're tanking.

You disagree with this, yet based on the results of your own test, the only time you got interrupted was during scenario 3, which you have admitted was flawed.

Quote:
2: Gearswap is 0s and 1s. All it sees, knows, and processes are current gear > pre cast gear > mid cast gear > after cast gear. It doesn't matter what we're doing, or what the original status was, it's logical processing is the same for all 3 scenarios. There is zero evidence to support that gearswap handles gearswaps differently based on the original status the command was triggered from. Input and output.

Not entirely sure what you are trying to say here?

Quote:
3&4: You're discussing macro spamming and inefficiencies. You have to understand the goal posts have been moved twice thus far.

Ultimately the same thing, whether it's spamming macro, or a single action, both can lead to errors, as previously proven.

Quote:
My test was to disprove that pre-cast and midcast are handled 'instantly' 'simultaneously' or 'same packet'. I proved there is a void between pre-cast and mid-cast in which a hit, check, and interrupt can occur. But, I likely created a flaw in the test, by capping the packet queue, so a new test needs to be done.

So, first I do the test, and they blame after-cast sets. I eliminate that variable in the test, now they blame the packet cap. Both are valid arguments and justify re-testing, they are correct.

I am curious how you intend to conduct this next test? I can't think of any other methodology to prove otherwise, therefore what is the point or purpose of further attempts, if we already know what causes the anomaly.

Scenario 3 right now is nothing more than a problem created from no problem, we have established the cause, and we know the remedy.

Quote:
Multiple people have stated, and I'm sure hundreds more have experienced, an interrupt when they didn't think it should be possible.

Vague and unsubstantiated, the facts remain visible. Also I would suggest you look at excellent written LUAs out there like Arislan's for example, and you certainly will not find a SIRD in precast, that's just absurd.

Lastly, and on a side note pertaining to SIRD, I would encourage players to sway away from relying heavily on SIRD sets all together. 2~3 years ago when I tanked on Paladin, I resorted to SIRD set in almost all situations, because it was the easy ticket out of spell interrupts, this came on the cost of potency. When I had a discussion with the Japanese LS about this habit, they very surprised to learn about that and told me that not a single Tank in their community relies on SIRD full time to land spells, and instead they time their actions properly and utilize vanilla's macro system, to equip FC and enmity set together with full potency, using a workaround which consists of the following command
Code
/grin motion <stnpc>
/equipset FC-Set
/ma "Spell" <lastst> <wait 1>
/equipset Potency-Set <wait 2>
/equipset Idle/Engaged

Sadly, there is a limit to the success rate of the spell landing in full potency depending how the spell duration and total FC%. (vanilla macros).

When I started playing RUN, I asked Thorny for advice on playing the job, and when I mentioned that I was planning to use SIRD full time the same way I did on Paladin, he advised me to consider that RUN has massive amount of FC, and that SIRD used in this manner is a 'crutch for poor timing', and that he is able to land spells without SIRD on full omen trash pulls without issues. At the time, this sounded very challenging to me, because I had yet to put this method into practical use, and once I did, I got use to it and never ever went back to SIRD on RUN ever again.
 Ragnarok.Lowen
Offline
Server: Ragnarok
Game: FFXI
user: Rorrick
Posts: 316
By Ragnarok.Lowen 2021-05-26 11:48:19
Link | Quote | Reply
 
There's not much reason to use SIRD for tanking 1-2 targets, but calling it a crutch for poor timing is also pretty reductive on his part. There are many situations where you're being hit by 4+ things where good timing becomes impractical or impossible. Not by coincidence, most of those situations are also the ones you're going to be /BLU for in the first place.
Offline
Posts: 18
By dennisnedry 2021-05-26 12:02:42
Link | Quote | Reply
 
Bahamut.Balduran said: »
When I had a discussion with the Japanese LS about this habit
Stopped reading right there. Japanese players arent dealing with across-the-ocean ping times. Japanese players were not impacted by the Odyssey lag that other players were very clearly being impacted by.
 Shiva.Thorny
Offline
Server: Shiva
Game: FFXI
user: Rairin
Posts: 2077
By Shiva.Thorny 2021-05-26 12:09:52
Link | Quote | Reply
 
Ragnarok.Lowen said: »
There's not much reason to use SIRD for tanking 1-2 targets, but calling it a crutch for poor timing is also pretty reductive on his part. There are many situations where you're being hit by 4+ things where good timing becomes impractical or impossible. Not by coincidence, most of those situations are also the ones you're going to be /BLU for in the first place.

If you're training mobs, just get a little distance and then run straight into them. They all swing as soon as you enter range, then they're synced and you have 4 seconds between rounds to cast.

If you're holding hate on adds or multiple mobs for a longer term fight, using cureskin will make it easy to get casts off in full enmity(especially if you dualbox a mage and aren't botting it).

There are certainly times where it's useful, but far more often you're just reducing your overall performance because you don't want to put the extra effort into timing.
[+]
 Ragnarok.Martel
Offline
Server: Ragnarok
Game: FFXI
Posts: 2894
By Ragnarok.Martel 2021-05-26 12:13:55
Link | Quote | Reply
 
Personally, I don't really use SIRD On RUN either.

But for PLD I think there's something to be said for not having to wait to cast. 1 second of waiting here, 2 seconds there. Eventually all the waiting to time spells adds up to extra casts you could have had if you didn't have to wait. Although that is somewhat offset by the reduced enmity+ on SIRD casts.
[+]
 Asura.Cambion
Offline
Server: Asura
Game: FFXI
user: Cambion
Posts: 415
By Asura.Cambion 2021-05-26 13:32:23
Link | Quote | Reply
 
Bahamut.Balduran said: »
Stuff

You seem to be extremely confused about the discussion at hand. I think you need to re-read the last 2 pages and re-*** how the conversation has evolved.

Your conclusion and statements thus far, aren't based on reality. You're pretending that Scenario 1 and 2 don't exist, based purely on the fact that because the original status literally can't be interrupted, that they wouldn't be. That's not a scientifically sound assumption.

The test will be repeated at macro intervals that don't cap the packets. As was already explained above.

One of the current main activities, Odyssey, requires Pld to super tank ~30 mobs, repeatedly for 30 minutes non stop. Trying to 'time' your spells and not having a SIRD set, again, is not based on reality. I'm not risking the efficiency of my groups run on 'timing' or special tactics to sync attack animations.

You're still talking about macro mashing, and completely misinterpreting the means of expedited testing vs 'the cause' of what's being tested. We are not talking about Macro Mashing. This was just a way to conduct the test to speed up what was actually being tested/theorized.
Dane was completely accurate when he stated multi-macro pushes, will result in after-cast gear sets, which I already tested/agreed would absolutely result in interruptions that have nothing to do with Pre-Cast.

Let me try to simplify this, so that we can just move forward, and dismiss the prior conversations or claims, that I didn't originally make in the first place.

My line of questioning is as such:

1: Does your character ever physically equip Pre-Cast gear in game? Yes or No?

2: In that split second of Pre-Cast gear being equipped, are you starting the spell cast? Yes or No?

3: In that split second of Pre-cast gear being equipped, can you be hit? Yes or No?

4: In that split second of Pre-cast gear being equipped, and getting hit, does the game to 'check' for Spell interruption rate? Yes or No?

5: If your SIRD in your Pre-Cast Gear isn't capped, can you be interrupted? Yes or No?

It's a very simple chain of thought. Whether it's applicable to coding, data transfer, and gearswap... is a different story.

How we have progressed so far.

1,2&3: So far, the responses have been 'no'. Claims of 'instant' and 'simultaneously' were made stating they are handled in the same packet and it's not a linear 'time' scale, but a 'point' in which that cannot be checked by the game.

Okay, so if the theory is that the packets are handled instantly and simultaneously, it shouldn't matter how often I hit my macro, because pre-cast gear is never actually occurring, right?

Test Results: Interrupted.

Response, "this is caused by after cast, not pre cast'

Deleted all aftercast from lua. The only 2 sets remaining are pre cast and mid cast.

Test Results: Interrupted.

Response, "you sent too many packets, that's the only reason precast gear was ever equipped/checked/interrupted."

So, next is to repeat the test, on a slower interval, so that packet cap is no longer a variable. Results to be determined.

You have to understand, that my current theory is like... 1/100,000 chance of even happening in a normal game atmosphere maybe?
Even if we assume my theory(not even theory, it's just pure curiosity) is correct, the .01 seconds that Pre-cast is equipped, in order to get 'hit' in that time frame, is probably 1/10,000 chance with a single mob? The timing would have to be so incredibly unlucky.
Then you take that 1 hit, and you would have to repeat it enough times to see an assumed 9/10 interrupt rate?

This is why the macro spamming was used to try and mimmick as many of these milli-second scenarios as possible. Of course, this was prior to understanding the packet limits, which invalidated the test, therefor the need to re-test.

I was not testing macro mashing. We universally agree that's terrible.
I was using macro mashing, to replicate this split second, as many times as I could, while being hit by as many mobs as I could, to try and replicate a scenario that may, or may not, be possible. It's important not to confuse these 2.

Honestly, a more helpful question, would be... How different is the timing in game/gearswap for a pre-cast spell vs a weaponskill?
Are they handled differently? Is the duration of time in which gear is equipped different? Do you believe the packets are handled the same way?
Logically, we've all been smacked while in our WS gear, so the scenario is similar and comparable in theory, but only if the smart people say they're even handled the same from a coding/packet point of view. If not, it's a moot point, but my curiosity comes from the same logic of that handling. Instead of an interrupt, we can visually see increased damage, and we know advanced luas for tanking, DO balance DT in their WS sets.
 Phoenix.Mikumaru
Offline
Server: Phoenix
Game: FFXI
user: Mikumaru
Posts: 382
By Phoenix.Mikumaru 2021-05-26 14:47:37
Link | Quote | Reply
 
Asura.Cambion said: »
Bahamut.Balduran said: »
Stuff

You seem to be extremely confused about the discussion at hand. I think you need to re-read the last 2 pages and re-*** how the conversation has evolved.

Your conclusion and statements thus far, aren't based on reality. You're pretending that Scenario 1 and 2 don't exist, based purely on the fact that because the original status literally can't be interrupted, that they wouldn't be. That's not a scientifically sound assumption.

The test will be repeated at macro intervals that don't cap the packets. As was already explained above.

One of the current main activities, Odyssey, requires Pld to super tank ~30 mobs, repeatedly for 30 minutes non stop. Trying to 'time' your spells and not having a SIRD set, again, is not based on reality. I'm not risking the efficiency of my groups run on 'timing' or special tactics to sync attack animations.

You're still talking about macro mashing, and completely misinterpreting the means of expedited testing vs 'the cause' of what's being tested. We are not talking about Macro Mashing. This was just a way to conduct the test to speed up what was actually being tested/theorized.
Dane was completely accurate when he stated multi-macro pushes, will result in after-cast gear sets, which I already tested/agreed would absolutely result in interruptions that have nothing to do with Pre-Cast.

Let me try to simplify this, so that we can just move forward, and dismiss the prior conversations or claims, that I didn't originally make in the first place.

My line of questioning is as such:

1: Does your character ever physically equip Pre-Cast gear in game? Yes or No?

2: In that split second of Pre-Cast gear being equipped, are you starting the spell cast? Yes or No?

3: In that split second of Pre-cast gear being equipped, can you be hit? Yes or No?

4: In that split second of Pre-cast gear being equipped, and getting hit, does the game to 'check' for Spell interruption rate? Yes or No?

5: If your SIRD in your Pre-Cast Gear isn't capped, can you be interrupted? Yes or No?

It's a very simple chain of thought. Whether it's applicable to coding, data transfer, and gearswap... is a different story.

How we have progressed so far.

1,2&3: So far, the responses have been 'no'. Claims of 'instant' and 'simultaneously' were made stating they are handled in the same packet and it's not a linear 'time' scale, but a 'point' in which that cannot be checked by the game.

Okay, so if the theory is that the packets are handled instantly and simultaneously, it shouldn't matter how often I hit my macro, because pre-cast gear is never actually occurring, right?

Test Results: Interrupted.

Response, "this is caused by after cast, not pre cast'

Deleted all aftercast from lua. The only 2 sets remaining are pre cast and mid cast.

Test Results: Interrupted.

Response, "you sent too many packets, that's the only reason precast gear was ever equipped/checked/interrupted."

So, next is to repeat the test, on a slower interval, so that packet cap is no longer a variable. Results to be determined.

You have to understand, that my current theory is like... 1/100,000 chance of even happening in a normal game atmosphere maybe?
Even if we assume my theory(not even theory, it's just pure curiosity) is correct, the .01 seconds that Pre-cast is equipped, in order to get 'hit' in that time frame, is probably 1/10,000 chance with a single mob? The timing would have to be so incredibly unlucky.
Then you take that 1 hit, and you would have to repeat it enough times to see an assumed 9/10 interrupt rate?

This is why the macro spamming was used to try and mimmick as many of these milli-second scenarios as possible. Of course, this was prior to understanding the packet limits, which invalidated the test, therefor the need to re-test.

I was not testing macro mashing. We universally agree that's terrible.
I was using macro mashing, to replicate this split second, as many times as I could, while being hit by as many mobs as I could, to try and replicate a scenario that may, or may not, be possible. It's important not to confuse these 2.

Honestly, a more helpful question, would be... How different is the timing in game/gearswap for a pre-cast spell vs a weaponskill?
Are they handled differently? Is the duration of time in which gear is equipped different? Do you believe the packets are handled the same way?
Logically, we've all been smacked while in our WS gear, so the scenario is similar and comparable in theory, but only if the smart people say they're even handled the same from a coding/packet point of view. If not, it's a moot point, but my curiosity comes from the same logic of that handling. Instead of an interrupt, we can visually see increased damage, and we know advanced luas for tanking, DO balance DT in their WS sets.
The best practice is trial and error. Square doesnt want repeatable results on the hardest content. They have been saying since the days of Absolute Virtue there's an element of "luck". Its why Odyssey Gaol has the mechanics it has. You'll have play with things and see what gets you the best results.
You'd have to do your test in an instanced area such as Omen/Ody/Dyna. Testing in a dungeon or Abyssea or Escha isnt the same result as those 3 areas.
Offline
Posts: 1109
By DaneBlood 2021-05-26 15:12:27
Link | Quote | Reply
 
Asura.Cambion said: »
1: Does your character ever physically equip Pre-Cast gear in game? Yes or No?

2: In that split second of Pre-Cast gear being equipped, are you starting the spell cast? Yes or No?

3: In that split second of Pre-cast gear being equipped, can you be hit? Yes or No?


1+2+3
This is what i have been trying to explain to you. you keep thinking real world. with a close to infinity resolution of time. that is not how it works in a game.

the current belive is that in the same "split second" on the small nucloid of time units. it all happens at that same time unit.
whatever you want to call it a year a minutte or a split second does not change that its it the small amount of units in the game that can be switched around. aka nothing else in the game is happening at the same time.



it complete the same as posting on this forum
I can write

PRECAST GEAR
SPELLCAST
MIDTCAST GEAR

and non amount of you hitting he submitt button will have anything from you posting happening in between those lines. because theay are in the same group of time units in forum posting.

Thet are clearly still in order but nothing can happen to come in between my lines written above

It is really no different
Offline
Posts: 1109
By DaneBlood 2021-05-26 15:15:13
Link | Quote | Reply
 
no im writing
AFTERCAST

could a post have come in between midtcast and after cast ( and therby interupted me?)
yes offcause because it 2 differen post of information.. it is 2 different packets.
They are happeninh at 2 different time units and something has been checked in between.
 Asura.Cambion
Offline
Server: Asura
Game: FFXI
user: Cambion
Posts: 415
By Asura.Cambion 2021-05-26 15:24:34
Link | Quote | Reply
 
DaneBlood said: »

PRECAST GEAR
SPELLCAST
MIDTCAST GEAR

Thet are clearly still in order but nothing can happen to come in between my lines written above

Counter Argument:

WEAPON SKILL GEAR
WEAPONSKILL
TP GEAR
"Nothing can happen to come in between my lines written above"

And yet... we've all been hit by a mob in our TP gear.

So, are you saying precast and spells work completely different than Weaponskills? Or are the packets sent completely different between these 2? Why the disparity?
Offline
Posts: 1109
By DaneBlood 2021-05-26 15:25:28
Link | Quote | Reply
 
i would like to follow pu my previous post with a bit more detail on the 3 first question


1: Does your character ever physically equip Pre-Cast gear in game? Yes or No?
Nothing happens physically. its all data. if you want to get this Deep into it you need to defined what you consider "physically". Cause yeah it all data and calculations

2: In that split second of Pre-Cast gear being equipped, are you starting the spell cast? Yes or No?
Yes that is the current belief. i took the liberty to translate split second to time smallest possible context swicthing time unit

3: In that split second of Pre-cast gear being equipped, can you be hit? Yes or No?
I am again taking the liberty to translate split second.
No. nothing else happens IN THE ENTIRE GAME WORLD at least to some ressource/code limitations about the server software threading and the CPU quanta distribution schedule.
Offline
Posts: 1109
By DaneBlood 2021-05-26 15:30:02
Link | Quote | Reply
 
Asura.Cambion said: »

Counter Argument:

WEAPON SKILL GEAR
WEAPONSKILL
TP GEAR
"Nothing can happen to come in between my lines written above"

And yet... we've all been hit by a mob in our TP gear.

So, are you saying precast and spells work completely different than Weaponskills? Or are the packets sent completely different between these 2? Why the disparity?

You are absolut correct but its not a counter argument
If you send all the sets in the same package ( same post) than yes nothing can interupt it.

As you have been told precast and midtcast is send in the same package
It just happens that gearswap does not do that with aftercast because its waiting for the results of the cast.


Also please note term like TP and weapon skill gear are purpose termns. Precast and midtcast and aftercast are mechanical term.
Dont confusse the terms. As one determins when it are sent to the server the rest just whatever name you call it

-- edit to clarify:
Nothing stop you form putting you TP gear in midtcast and it will in fact go out in the same package.
but remember midtcast is for what affect the potency of the action.
So if you want to WS in TP gear ahead and put TP gear in midtcast

PRECAST happens right before the action that why we applky fastast cause the calcualtion is happening before the spell/action can start
MIDTCAST happens right after the command is set so we are in midtcast for when the results of the action is beeing calculated

AFTERCAST is what we go inte once the action is stopped

This shouldn't really be controversial. but im pretty sure at this point you are just trying to make a fuzz to make a fuzz rather than to understand.


TL:DR
TP gear is not in midtcast but in aftercast.
NOT the same package.
its that simple
 Shiva.Thorny
Offline
Server: Shiva
Game: FFXI
user: Rairin
Posts: 2077
By Shiva.Thorny 2021-05-26 15:31:55
Link | Quote | Reply
 
WS is not sent as one unit, there's a finite time from when your WS starts until it ends, you can even be stunned after readying but before finishing a WS. Again, I can only comment for AC and not GS, but I send the ws swaps prior to the WS packet(but in same block). I don't go back to idle until the server sends the WS complete packet, which amounts to a ~2 second interval.
[+]
Offline
Posts: 1109
By DaneBlood 2021-05-26 15:42:44
Link | Quote | Reply
 
@Asura.Cambion

Deleting a post because you realized you are wrong is a little cowardice..
Outline it with a proper redaction message is more helptfull in a debate as at least other will learn from the is take when reading.


Edit : Im *** blind


@Shiva.Thorny
Question
Hows does AC handle multiple actions request in the came. E.G
like casting teleport and then casting another spell in the casttime time of teleport.
does it still swap on the second action request or does it know not to do anything else until the spell from the first request is done ?

in otherwords. does AC have the same aftercast issues when button smashing that GS have ?
 Asura.Cambion
Offline
Server: Asura
Game: FFXI
user: Cambion
Posts: 415
By Asura.Cambion 2021-05-26 16:03:40
Link | Quote | Reply
 
DaneBlood said: »
@Asura.Cambion

Deleting a post because you realized you are wrong is a little cowardice..
Outline it with a proper redaction message is more helptfull in a debate as at least other will learn from the is take when reading.

What? I have deleted nothing. I'm literally in Abyssea try to do the test we're discussing.
 Shiva.Thorny
Offline
Server: Shiva
Game: FFXI
user: Rairin
Posts: 2077
By Shiva.Thorny 2021-05-26 16:08:01
Link | Quote | Reply
 
AC will swap on the second cast, but won't go back to idle afterwards. It sets a maximum delay for the spell to execute, then cuts the delay short when the spell complete packet is received. Idle is processed as a state, rather than part of the cast itself. So, spamming the same spell is fine, but spamming different spells would still mess you up(you'd end up in the set for the last spell you cast). So, no, aftercast will never be a problem in AC.

I also pulled up windower and threw together a quick test to see what their packet cap appeared to be. I queued manafont 50x, which is a 28 byte packet. According to packetviewer, 12 of them ended up grouped with the first outgoing 0x15, which is a 32 byte packet. Strangely, 38 of them ended up grouped with the second. I repeated the test with 200 manafont packets, and packetviewer showed only 6 of them being grouped with the first 0x15, while the remaining 194 were grouped with the second.

My takeaway from this is:
-Windower is not applying a static packet size cap. Either they are doing something that falsely reports injections when they have not happened yet, or they are discarding packets and claiming they were injected.
-Windower can limit packets as low as 227 bytes(a 7th would have made it 228, despite the actual chunk being 200, in the 6 packet example).
-A 227 byte packet will barely fit position update, precast, action and midcast at the same time. If either uses equip single instead of equipset, or something like itemizer goes off, or you chat in the same interval it is possible you could reach 228+ bytes.

I am not sure what would cause this discrepency, and suspect that without a windower dev clarifying, we will not know exactly. My best theory is that injected packets are prorated in some manner, so if you cast earlier in the interval more will make it into the sequence. I'm also not sure I can totally trust packetviewer, as I really do not believe 194 manafont packets could fit into compression at once. Perhaps they try to inject all packets that are more than one sequence ID old, and the game discards the rest.

Regardless, I think at this point you need to get a Windower developer involved to find out more about how exactly their packet injection works. These are all issues that are not present on Ashita, and by the looks of it, there's a solid case for your midcast NOT being on instantly, based on the understanding I can glean from these tests. Be aware that this is not because of a server limitation, and if that is the case, it can certainly be fixed on Windower's end.
[+]
 Bahamut.Balduran
Offline
Server: Bahamut
Game: FFXI
user: balduran
Posts: 270
By Bahamut.Balduran 2021-05-26 16:08:15
Link | Quote | Reply
 
dennisnedry said: »
Stopped reading right there. Japanese players arent dealing with across-the-ocean ping times. Japanese players were not impacted by the Odyssey lag that other players were very clearly being impacted by.

I'd like to point out those Japanese players (at least the ones I referenced) are still using the stone age vanilla macro system, so gimped FC sets there. I sympathize with players who face severe lag issues (specially certain locations in EU), but for the record, I am based in the Middle-East and have to deal with 'overseas' lag too, and gladly have zero issues not using SIRD to time my spells. And btw, West Coast players in the United States sometimes get better ping than I do. Go Figure!

Ragnarok.Lowen said: »
There's not much reason to use SIRD for tanking 1-2 targets, but calling it a crutch for poor timing is also pretty reductive on his part. There are many situations where you're being hit by 4+ things where good timing becomes impractical or impossible. Not by coincidence, most of those situations are also the ones you're going to be /BLU for in the first place.

Remember spells like Geist Wall and Poisonga are reduced to 0.6 and 0.4 seconds casting time respectively with 80%FC (now also moderately achievable on Paladin), topped with a decent Phalanx set, and perhaps popping Sentinel or Battuta, the chance of getting hit is greatly reduced, and your AOE spell cast duration become ridiculously fast, easily 4~5 packs on an omen Floor, and very low margin for error, it just keeps getting better the more you practice it, and becomes a natural habit.

Asura.Cambion said: »
So, are you saying precast and spells work completely different than Weaponskills?

Absolutely :)

As for why the disparity, perhaps the author of GearSwap is better suited to answer this question, but from what I've learned, think of the the precast>midcast scenario and remove the midcast equation, gearswap waits to receive communication that the WS landed until it puts you back in aftercast, and that's basically your window. Edit: Thorny explained it much better in his previous second post.

Also, in response to your previous long message, although there few disagreement, I won't discourage you to run more tests, and would be glad to view the findings with you.
Offline
Posts: 1109
By DaneBlood 2021-05-26 16:11:17
Link | Quote | Reply
 
@Asura.Cambion

I will move forward and try to explain it a bit more

WS does not have a fastcast phase, that is important to understand
When a was action is request it is started and then the efficiency is calculated a bit later and boom we do damage

with spells it is different. Spell have 2 differnt time they are calulated.
right before starting the casting time because we need to know how long this delay it. WE cant do it later as we would not know how long to wait for.
then again when the spell actuall take effect aka "lands" then at this point the game calculates the potency of the spell.

THIS IS THE ENTIRE REASON WE CAN HAVE FASTCAST AS A SEPERATE GEAR SET



PRECAST
Action start ( calculating timer before effect)
MIDTCAST
|||
<Undefined delay> ( this is where other things in the world are happening. aka interuption
|||
Action goes off (Calcualate potency of the action)
<unsure with time here>
AFTERCAST



As you can se we can send the precast and midtcast with the spell action start as we do not care about any time here. just that it is in order.
However we can send the after cast ( you tp set) as we dont know when the effect is actually have taken place and aftercast offcause needs to go AFTER the effect goes up (hence the name)

i tink you might be confussing the term midtcast as something that gest equip when the spell is going off. it is not its is sonething that HAS to have been equip when the spell goes off. we cna put it in long before the spell goes off. heck even in precast if we wanted to as long as there are no change in midtcast. that is why it can happen in the exact same time unit that the spell going off happens and therfore before any interupts can happen

This is the current belief/understanding on how gear swap is working. and te reason to why "We" believe there is no time between precast and midtcast to be interupted in
Offline
Posts: 1109
By DaneBlood 2021-05-26 16:13:51
Link | Quote | Reply
 
Asura.Cambion said: »
DaneBlood said: »
@Asura.Cambion

Deleting a post because you realized you are wrong is a little cowardice..
Outline it with a proper redaction message is more helptfull in a debate as at least other will learn from the is take when reading.

What? I have deleted nothing. I'm literally in Abyssea try to do the test we're discussing.

Yeah i was going blind.
my apologies
Offline
Posts: 1109
By DaneBlood 2021-05-26 16:15:17
Link | Quote | Reply
 
Shiva.Thorny said: »
AC will swap on the second cast, but won't go back to idle afterwards.

But you will be in the midtcast set for the second action?
If i am reading this correctly.
 Shiva.Thorny
Offline
Server: Shiva
Game: FFXI
user: Rairin
Posts: 2077
By Shiva.Thorny 2021-05-26 16:15:33
Link | Quote | Reply
 
DaneBlood said: »
But you will be in the midtcast set for the second action?
If i am reading this correctly.
Yes, that is correct. In your teleport > barspell example, you would see:

Packet chunk 1:
teleport precast
teleport
teleport midcast

packet chunk 2:
barspell precast
barspell
barspell midcast

packet chunk x(when teleport finishes casting):
back to idle

may be worth adjusting the logic for ashita4, to prevent the barspell from equipping at all, but there's a point where you have to make some sacrifices(if you don't equip midcast until you confirm the cast started you're likely too late for many spells.. so you realistically need to trust the user at least a little bit)

however, hitting the same spell any number of times will have no ill effects, as you'll just be sitting in that spell's midcast after each instance and won't go back to idle until it actually completes
[+]
 Asura.Cambion
Offline
Server: Asura
Game: FFXI
user: Cambion
Posts: 415
By Asura.Cambion 2021-05-26 18:00:33
Link | Quote | Reply
 
Okay, well I'm tired of teleporting myself out of abyssea.

Test was conducted as such:
Pld/Whm, with Protect V, Regen Atma, Refresh Atma.
Lua was limited to a Precast set(full DT, 10SIRD from merits) and a Midcast set (115% SIRD) This results in only 2 sets possible, either DT or SIRD, no aftercast is ever taking place.

I gathered all of the mandies in Abyssea-Altep
Cast Teleport <whatever>
I then re-cast a new spell (phalanx) at every 5% of the spell casting.
So I hit my Phalanx Macro at 5%, 10%, 15%, 20, all the way to 55%, and then manually input /heal to cancel the teleport.
Allowing for human error, response time, laziness, I averaged 9-10 cast attempts per teleport cast.
This is likely 3 attempts per minute, resulting in 30 attempted casts per minute.
I did some back and forths because the Teleport went off at least 3 times, I ran out of mana once and died, lol, and tried chigoes in aby attowa the final time. With /heal tracking my time remaining in abyssea, I was in Attowa for 25 minutes, minus teleport, mob gather etc, but combined with time in Aby Altep, I'm confident I was testing this for at minimum 30 minutes.

This results in 900 cast attempts at a minimum in my estimation. And in those 900 attempts, I was not interrupted a single time.

This leads me to believe that the packet-capacity of my last test is what caused the interrupt, not the Pre-Cast gear.

Conclusions:
1: For normal Luas, your interrupt issues are caused by your 'aftercast' handling. I removed this variable from my test, as it was simple to confirm that aftercast will create problems for people who hit their macro more than once.
Example: Macro hit > Pre cast > Mid Cast > Macro hit > Pre-cast > (fails) > returns to aftercast/idle/tp set = Your midcast gear that has all your SIRD is NOT on, and you WILL get interrupted.
*This is a flaw in Gearswap in my opinion, as a status check should always return that I'm in a casting state, not idle state, but I digress.

2: Pre-Cast sets, do not appear to require SIRD, as they can not be interrupted. People far smarter than me explained it above, but Tl;dr it has to do with how the game handles 'packets'. Gearswap sends the precast, the spell, and the midcast all to the system at once.
*I'm going to elaborate here for future people like me.
2a) The gear is still equipped on the back end, and despite explanations above, it still happens over an instance of time, as far as chain commands exist in computers. However, it's only happening in 'coding time'. What that means is, if you had a precast set with a hypothetical 1hp build, and a midcast set with 1000 hp build, your character HP will drop to 1hp, as the game processes each gear change on the computing side. BUT... it never happened on the game side, therefor nothing in-game can interact with your character during those computed changes/commands, because no time has passed "in game" between these commands that happened "while computed". That's how my brain processes that anomaly. (Even if not technically accurate, the point is to explain to like minded people)
2b) If you logically, like me, assumed that WS Gear and Pre-Cast gear would function the same way... They don't. In our heads and our gearswap we program them the same, but the way the packets are handled on the back end is completely different. This is why 'in game' the chain of gear changes CAN be effected by your surroundings/mobs you're fighting etc, but pre-cast can not.
2c) There is a limit to packets, that my previous test was able to exploit. I wasn't aware at the time, but tl;dr you can force an interrupt between pre-cast and mid-cast, by sending so many gear-swap requests to the server, that it can only process part of them, resulting in only part of your gear being applied within the 'instance' of data packet sent to the server, allowing your character to be effected by the result, as your request was so large, it had to happen over 2 instances of data transfer to the server.
3) I said this even before I began my tests, but even if you could get interrupted by the hypothetical split second in pre-cast, it wouldn't be worth it, as you would have to sacrifice far too much, for something that would so rarely ever occur.
4) I didn't start this debate, someone else did, but I took their thesis and ran with it because I wasn't logically able to explain why it couldn't be possible and the first test I did seemed to contradict the existing theory. I was curious. I was wrong.

tl;dr
Don't multi-hit macros. Don't worry about SIRD in precast.
[+]
Offline
Posts: 1109
By DaneBlood 2021-05-26 20:43:49
Link | Quote | Reply
 
Asura.Cambion said: »
Okay, well I'm tired of teleporting myself out of abyssea.

Didn;t have time and the mental capacity with me right now to read.
Just wanted to give a cudos for the testing.

I'll give my thought later
 Pandemonium.Zeto
Offline
Server: Pandemonium
Game: FFXI
Posts: 368
By Pandemonium.Zeto 2021-05-27 00:34:24
Link | Quote | Reply
 
Asura.Cambion said: »
1: For normal Luas, your interrupt issues are caused by your 'aftercast' handling. I removed this variable from my test, as it was simple to confirm that aftercast will create problems for people who hit their macro more than once.
Example: Macro hit > Pre cast > Mid Cast > Macro hit > Pre-cast > (fails) > returns to aftercast/idle/tp set = Your midcast gear that has all your SIRD is NOT on, and you WILL get interrupted.
*This is a flaw in Gearswap in my opinion, as a status check should always return that I'm in a casting state, not idle state, but I digress.
If macro spamming is an issue, you should be able to add a quick cancel_spell() check. I'm not sure if you can get if your actual casting state/the code for it but you can start your precasts with a basic concurrency lock and if the lock is taken, cancel_spell.
Code
if lock then
cancel_spell()
end
lock = true
<precast>


Then in your aftercast
Code
lock = false
<aftercast>


And then give lock proper scope in your file. You may want an emergency unlock bind because issues
[+]
Offline
Posts: 1109
By DaneBlood 2021-05-27 10:01:19
Link | Quote | Reply
 
@Asura.Cambion


This leads me to believe that the packet-capacity of my last test is what caused the interrupt, not the Pre-Cast gear.
Wooot SCIENCE.
I Agree with this conclussion (until newer evidence arrives)

*This is a flaw in Gearswap in my opinion, as a status check should always return that I'm in a casting state, not idle state, but I digress.

This is why at the very beginning in told that people should put in a cancel spell precheck in their lua to avoid the after cast gear.
It would not detect casting state but it would at least verify if a spammed spell is ready or not and not do gearswap if its not ready.

akai button smash one does it things but button smash 2 detect that the spell is not ready so cancel out and do nothing with gear,


but I took their thesis and ran with it because I wasn't logically able to explain why it couldn't be possible and the first test I did seemed to contradict the existing theory. I was curious. I was wrong.
You were not wrong. You had an incorrect theory.
When you do testing to verify/validate or disprove you theory. You are never wrong.
Wrong are people that have a belief with no evidence nor will to test it


Now you just need to go back and test it to confirm that aftercast with no SIRD will make lots of interupts when button spamming :D
Im joking.
Big thank you for doing the testing.
[+]
necroskull Necro Bump Detected! [72 days between previous and next post]
 Asura.Otomis
Offline
Server: Asura
Game: FFXI
Posts: 164
By Asura.Otomis 2021-08-07 16:43:13
Link | Quote | Reply
 
Handy Resource:

https://kanican.livejournal.com/20380.html
[+]
Offline
Posts: 1584
By Felgarr 2021-08-07 18:24:02
Link | Quote | Reply
 
What do the C's and D's mean for certain spells?
 Siren.Kyte
Offline
Server: Siren
Game: FFXI
Posts: 3331
By Siren.Kyte 2021-08-07 18:36:37
Link | Quote | Reply
 
Cure based, damage based
First Page 2 3 4 5 6 7 8