Monday, August 23, 2010

Achievement Unlocked

One of the things on my "long-term unlikely but hope to get items list" was RDM relic headgear: Duelist's Chapeau. I can now scrap that item off the list because as of last night I'm the owner of my very own "Relic Pimp Hat". As if that weren't enough, I get me pair of these:

That's two highly sought-after Dynamis drops in one run. Needless to say I was and AM very happy. In both cases I was mentally preparing myself to NOT get the item. I'm usually not at the very top of the attendance list, especially not lately with Abyssea runs which I don't attend. I was taken aback by all the congratulations I received from fellow LS members. I feel like many people were genuinely happy for me for getting the drop(s) which was very touching. It's going to sound corny and/or cheesy, but that was probably more meaningful than the item itself.

As for practicality, up until now I've been using Nashira Turban for my enfeebles. It has the same amount of magic accuracy as the RDM JSE Wise Cap, but with the added benefit of -5 Enmity, 2% haste. I suppose the -10% Spell Interruption does account for something as well. Compared to Nashira, Chapeau has +10% accuracy rate on almost anything (where your base accuracy is above 50%). Read this for a better and more complete description of how magic accuracy works.

Aside from making my Idle set look less bright (I shall miss you, V.Cloak), D.Chap increases my enfeebling magic accuracy by 10% on a very large number of NM's... until capped. So right now I don't know how far from capped accuracy I would be on a random NM like Faust. I have a very decent set for enfeebling magic already as well as capped enfeebling merits and 4 Ice Accuracy. In theory, when casting Bind on Faust while soloing (a solo I have never done) the 2% Haste on Nashira would outperform D.Chapeau if my accuracy was capped. Does that mean I'll keep using Nashira for Bind on NM's? The answer is "No". Bind is a pretty essential spell when solo'ing. If having it resisted is not an option, then assuming you're at capped accuracy is too big of a risk in my personal opinion. Nashira still has a place in Dark magic macro's and as such won't be put into storage. My Vermillion Cloak however, porbably won't see any action unless I decide to level BLM or SMN any further. It's time I reworked my macros.

Friday, August 13, 2010

WoE attempts 1 and 2

Last Thursday we gave WoE our first attempt as a linkshell. Based on info from wiki and BG I devised a rather straightforward strategy. Killing the first crab and running away seemed simple enough. However, not accounting for random DC's got the better of us. One person in alliance DC'd while the runners got close and we weren't able to dissolve and disband before they aggro'd. I tried to salvage it by getting melees to kill the new 3 crabs simultaneously. If we managed to do that we could avoid aggro from the next wave of runners by getting out of the way. It was a bit too chaotic to explain this at the time so it didn't happen. After one crab died a boss came in the wave of runners and I told everyone to wipe asap.

After getting back up we proceeded with the rest of the plan. Tillaert and Muze did an excellent job of sac pulling the Caldera Crabs. Muze, on PLD had to survive with 8+ crabs beating on her while we got a Caldera Crab off the train. The Caldera stopped to cast a spell on the person pulling off the train and so it wasn't entirely foolproof. If the sac'er dies before that time, they're able to re-link immediately according to what I've read.

We started off with a pair of MNK DD tanks on the boss but the BLM's pulled hate pretty much immediately and one of the MNK's went down pretty fast. Not sure what it was though I suspect it was Mega Scissors. I should have recorded it. Anyway, Argonar said he was hitting for 0's. Probably the result of the 1k Stoneskin it can get from the Metallic Body ability. Again, I really should have recorded it... The BLM's went down and Argo couldn't keep hate while not doing any reasonable damage so the attempt ended prematurely.

I treated it a bit like the VNM Krabkatoa, but these seem tougher to deal with to a certain extent. We brought PLD's for the second attempt. The first crab kill went according to plan this time. I suppose in the future we could do this while everyone is solo and form alliance while the sac'ers take up position. This, just in case someone's too late running away from deathspot. They can then die without it requiring a full wipe. The Caldera was pulled in the same way and we started off well. Hate was much more stable after asking BLM's to hold back for a while. However, the hate reduction from Mega Scissors is very strong. Even with PLD's, the BLM's pulled hate off them even after letting them build up hate beforehand. I realise a BLM could do this regardless, but it happened sooner than expected. Argonar mentioned it did Mega Scissors several times in a row at one point. Man... I really should have recorded it...

Around 50% the tank party took some casualties (RDM + WHM) due to the Caldera being out of position and taking damage from Mega Scissors as a result. We weren't able to recover and lost. In addition to hate not being manageable, it has a lot of HP and takes reduced damage from melee. BLM's seemed to do well, but they're very fragile.

Several people have suggested to me that using a skillchain and magic bursting would let us take it down faster. I agree the damage would be much better and there's the chance we could expose a weakness and make it easier. The second alternative is to take it down with melee, but with a lot more power and numbers. You could treat it like a zerg that you have to repeat 3 times. The challenge for both is staying alive long enough to kill it. A stun rotation would help but a Venom Shower is sure to get through. If you rotate bards like it were Einherjar you can buff melee with Carol, Barwater and Paeon, and still get some damage enhancing songs. Water resist seems to work well even without any additional resist from gear. The accuracy of Venom Shower isn't floored, but it's enough not to worry about it and focus on damage taken from Mega Scissors. There are bound to be deaths, which would cut it pretty close if you take more than 3 mintues to kill a Caldera Crab.

Monday, August 9, 2010

Weekend update 8

Magian trials can be really tedious. I'm now halfway through the last trial for my Store TP Keito and it's taking a long time. Efunotachi is stuck on the very first Opo-opo trial due to the non-existance of fire weather in Jungles/Cauldron. I regularly tele out to Kazham to see what's coming up, but even in the "summer" months of FFXI fire weather would mostly show up unannounced. I only need 7 more kills.

Besides some work on trials I got two levels on DRK. It's been awful really. Not a lot of people seeking outside of the subjob level range and 75+. I got an invite for an EXP party in Crawler's Nest. I took it cus I was desperate. It was about the same rate of EXP I get doing campaign. So afterwards I opted for the latter. Campaign is still popular outside of Northlands on Valefor. We haven't even controlled Beacedine for at least two weeks. The restof the zones are usually full of people, especially Windust zones. After a while I get an invite for an MMM party around level 27. It took a while to get started, but the setup was pretty much perfect and I got a nice bump from the two runs I joined. I couldn't stay for the rest, which was very disappointing. I had Nyzul to do.

Goliard Saio and Askar Zuchetto dropped from the 4 runs. Really want that Askar Korazin. Since I have more tags than the rest, I could probably get away with doing some pickup parties during the week. Using SCH sub in Nyzul has been fun. I can toss everyone a Protect5 and Shell4 at the start without using half my MP and still have RR spell and most status removal spells. Accession Stoneskin is a nice bonus. The only thing that's missing is Accession working on Haste. I still don't understand why they stopped that from happening. Erasega was nice when we got Flayers.

Sunday we got revenge for the Valkurm loss. Other than Miasmic Breath at the start that went off, I don't think it did anything threatening. That poison is pretty ridiculous though. No time to cure myself while stunning I was anxious to see it go down fast, which fortunately it did. In contrast to previous Valkurm runs we killed fast enough to kill all Hippogryphs, Manticores on south-east beach as well as the Sabotenders near oasis.

Friday, August 6, 2010

Windower advantages

Avesta solo'd Byakko recently. There's a link to the video in that thread. What you'll also find in that thread is a heated argument (if you can call it that) about whether or not using Windower gives you an advantage and if it's an unfair advantage as well as whether or not Avesta's solos take more skill than if he used Windower and Spellcast.

Well, first off I can tell you that I have no personal experience with Spellcast. I've seen snippets of scripts that give me a broad idea of what it's capable of. I've used Windower a long time ago. At the time it was so I could dualbox and see my EXP/hr in parties. I haven't used it since SE made their own windowed mode. I don't use it, because
  1. I'm paranoid about being banned for it, despite all the evidence that it won't happen.
  2. I believe, in the majority of cases, the game should be played with the means the developers provided and nothing else.
Given the Byakko solo as well as similar fights I can think of the following things third party tools could have done to make it easier:
  • More efficient gear swaps
  • Distance plugin
  • On-screen recasts
  • More distance gained from JA0Wait (not windower related, but still 3rd party)
Avesta did not seem to use multi-layered macros that span multiple sets. I personally do this because I don't want to limit the number of slots I swap out. This introduces new problems though. First of all, to be efficient you have to start on the same set everytime. I use a macro to send me back to the first set in every spot that I don't use. As long as you control your button presses and you don't panic, you can do the same things with regular macros as you can with Windower macros. If, for example, I have my Bind macros across two sets, then I can use Fastcast in one and swap back the magic accuracy/skill gear after the magic cast line in the second. The benefit is mostly in being able to put more precisely timed pauses as well as eliminating the chance that you start your macro on the wrong set.

A secondary benefit is the small time gained by only having to press a button once. Some people make this point laughable by claiming that pressing a button 3 times doesn't take more skill than pressing it once. The point is that it takes more time and, as mentioned before, is more prone to error. You can not spam the macro presses. From someone who has always relied on intricate in-game macros, I can tell you that pressing the macro buttons too fast in sequence can cause some equipment swap lines to be ignored/skipped. Recently my idle-set macros have gone from two sets to three. I now have to press ctrl+1 three times to go back to my neutral gear after taking an action. I admit that I get sloppy when doing this and often don't go back to my idle set between actions because I want to be able to do things in quick sequence. This is a non-issue when using spellcast gearswaps, which are more powerful than Windower macros.

Against any mob whose magic or abilities can be avoided by standing at the correct range, the distance plug-in allows you to do this with a lot more precision. Sometimes there is a small band you can stand in where you're both safe from an attack and able to cast on the mob and/or tanks. I personally find this simple to figure out when fighting something in a group. I just take a few seconds to find the maximum casting range and I try to stay there. In certain solo's however, exact positioning can be a big benefit. Seconds are lost when you try to cast a spell on a bound or otherwise immobile mob, only to find out you need to take a few steps back to reach. Will this be the determining factor in the solo? Unlikely, but as I learned from my attempts at the Requiem of Sin battle, it does certainly matter.

Knowing whether or not a spell or ability is up, or knowing how much time is left on a buff before it wears off makes any battle more efficient. Since SE added the recast command I try to use it as much as I can. However, I often end up removing it simply because it takes up macro line space I'd rather use for something else. More ofte it's because I find it impractical to put a recast line in a macro of the spell I'm trying to cast (barring Refresh and Haste). In practice I would try to cast something, find out I can't yet, then go back to my idle set before I try again, find out I'm 1 second too early and I cycle through the macros sets again until the spell comes out. Blinking aside, it also introduces wasted keypresses which could have been used for something else. Why not use a specialized recast macro? Valid question. My answer to this is that the macro I would make would have at least 3 recasts in it. Visually parsing this moving list of numbers takes time. Also, the chat log updates slowly. By the time I've figured out what the actual recast is I may have pressed the macro three times due to chat-log spam from other things going on. In the end I find it not to be worth the precious macro real-estate and live/deal with sometimes pressing a macro too early. It's not as big a problem for my personal style of play. In cases where it could be, I use the recast macro command.

Though the distance gained from blink casting is not extreme, there are some clear examples of where this small difference is essential. One of them is Kaeko's impressive Bhaflau Remnants Solo where a single hit form Long-bowed Chariot could have ended the attempt. It also made his Kaizer Behemoth solo go notably faster. The argument that this can be done manually is somewhat invalidated by the fact that lag and response time as well as computer speed make this hard to do reliably. Windower macros can be helpful here too due to its more precise delay mechanism. I came up with a way to blink out Blizzard3 about 75% of the time using a sequence of 3 different /wait commands which made it so I could delay for a fraction longer than the sum of all the waits. Once I started doing this soloing Iriz Ima went much faster. It didn't make me take less damage, since, even with the spell-completion freeze, it's too slow to catch up as long as I start moving right before the spell completes. What it DID do was allow me to spam spells much more frequently as I could widen the gap between me and the NM much more quickly.

For all the benefits of these 3rd party tools, each can be argued to be only a marginal improvement. I believe that with proper planning, preparations, strategy and player skill, the difference these tools make is LESS essential to success than the race of the character. Galka vs Taru MP pool for example. However, be they marginal or essential (in certain specific situations) they add up and improve efficiency and as a consequence, the overall chance of success. I can't think of a solo right now that REQUIRES these tools to achieve. So, as far as I know, right now any solo done with them can be done without them. However, to me, claiming that these do not improve the odds of success sounds like a defensive reaction from people that do use them, rather than a balanced well evaluated opinion.

Wednesday, August 4, 2010

Great Katana options

Some people in the FFXI community have put a lot of work in trying to figure out all the math behind damage calculations. Some of it is over my head, so I keep to my own simplified logic which I then try to rhyme with what others are saying. Looking for the most ideal Great Katana to replace Hagun for my SAM, I tried to rely on my own formula.

DoT =
(WS_Rate * Average_Melee_Damage + ACC_Adjusted_Average_WS_Damage) /
(Actual_Delay * WS_Rate/Melee_ACC) + 2

Where WS_Rate is how many melee attacks are performed for each WS. This, is basically the make or break factor in my calculations. Although I'm confident that my estimation fits my personal situation (frequency of meditate usage and efficiency of TP usage), it may require some adjusting as I'll discuss later.

The idea behind the above formula is adding the melee damage and WS damage together in proportion to the WS_Rate and divide it by the time it takes per WS in seconds. You may notice the constant number 2 in there. From recording multiple WS and counting melee swings and the time it took, I've concluded that WS and ability delays are both approximately 120 delay or 2 seconds long.

The ACC_Adjusted_Average_WS_Damage may also require some explanation. Basically, in my formula I calculate the expected average WS damage for when all hits connect. Multiplying this value by your WS accuracy rate accounts for missed hits, or in case of single hit weaponskills, a complete miss. This way the formula also applies to other weapon types and weaponskills. The WS accuracy rate is also modified by an accuracy bonus which applies to Kasha, Gekko and other weaponskills. In this case I added 50% acc rate and capped the total value at 95%. This is based on info from this BG thread.

Using the WS_Rate and dividing it by the melee accuracy rate I come up with the average number of attack rounds needed per WS (not rounded). The Actual_Delay is the weapon's delay adjusted with the Haste stat.

To come up with Average_Melee_Damage I used the following.

(Base_damage + fSTR) * (Average_PDIF + MIN(1, 3 - Average_PDIF) * Crit_rate)

The contribution of critical hits to your average damage is dependant on your average PDIF. Normally you'd add 1 to your PDIF for a critical hit, but when your PDIF is 2 or higher the actual contribution is reduced since PDIF is capped at about 3. That explains the MIN function. The question of the origin of my fSTR and PDIF numbers can be answered by: I used estimates from parsed fights on Seaboard Vultures. So, it's not precise like some other people would prefer, but it's good enough for my own ends.

For average WS damage.

(WS_Base_Damage * fTP + WS_Base_Damage * (Additional_WS_hits + Multi_Attack_Rate)) * (Actual_WS_PDIF + MIN(1, 3-Actual_WS_PDIF) * WS_Crit_Rate) * WS_Damage_Bonus


WS_Base_Damage = Base_damage + WS_fSTR + WSC


Actual_WS_PDIF = MIN(2.7, PDIF_During_WS * WS_PDIF_Bonus)

Since Gekko can't crit naturally (i.e. without Sneak Attack) the WS_Crit_Rate is going to be 0 in all my comparissons. I added it for sake of completeness. I split up the formula to make it more readable, but I fear it may still be a bit confusing. I multiply the base WS damage by fTP for the first hit and add it again for all expected additional hits, which also includes Double and Triple Attack procs. The WS_PDIF_Bonus is the 2x multiplier for Tachi Gekko and 1.5 for Kasha. Other weaponskills also have this property.

WS_Damage_Bonus represents the boost from Overwhelm and other gear like the AMK hat. One has to take into consideration whether or not the bonus from the Kasha weapon, Overwhelm and other gear is in fact the same stat. If they are not, then the effects would be compounded. However, I assumed for this example that they are in fact all the same stat and so can be added up respectively.

Even though I got PDIF, accuracy and fSTR estimates from parses, I still needed to estimate the different values for the same variables for when I change gear in between the different setups I want to compare. For accuracy this is not much of an issue, but for PDIF and fSTR I used some creative estimates.

Haste is based on 50% Hasso usage and Haste spell. The more Haste present, the heavier any haste sacrifice from gear will weigh.

Last, but certainly not least. Calculating WS_Rate for different setups and weapons. I had a baseline of 4.4 WS_Rate from my parses. This obviously includes Meditate usage. I needed a way to calculate new values for WS_Rate for the different weapons and gearsets. This is probably the most questionable part of my DoT formula.

WS_Rate = Hits_to_WS / Avg_Attacks_per_round - 0.4

This approximates my measured WS_Rate given my 6-hit Hagun setup. Hits_to_WS for a 6-hit is 5, because you get TP from your WS. Subtracting 0.4 to account for Meditate is going to be skewed if your WS frequency goes very high or very low. The higher it is the less Meditate contributes and vice versa.

Avg_Attacks_per_round is what the name implies. However, care needs to be taken when dealing with both double, triple and "Occasionally attacks" weapon properties. I didn't automate this in the spreadsheet I used and did it by hand for the OAT Magian weapon.

And now for what you've all been waiting for!

The percentage score at the bottom of each column is the DoT relative to the DoT of teh Hagun 6-hit build I'm currently using. All the variable names on the left should be easy to pair up with the names I used in this post.

Notice how both the Keitonotachi weapon options perform worse than Hagun with my gear selection. The Kasha damage obviously loses quite badly given the low PDIF I had on birds. Level correction plays a big role here. On low level mobs where my PDIF could easily be above the 1.35 needed to cap it during WS, the Kasha damage weapon would outperform the Hagun build.

The 5-hit build's sacrifice in Haste and accuracy weakens it considerably. Since this is just for myself I didn't bother doing the numbers for a build including Usukane. It's worth noting here that this setup too will beat Hagun when accuracy is capped. When considering zergs, then the Haste sacrifice become more costly and would make this an unattractive option.

Keeping Hauby and Dusk Gloves on, the 5 Store TP on the Keito can make for a 6-hit build with the same gear as my current Hagun build, but with a Sword Strap. This plus the Strength+Attack Magian G.Katana seem very close. This is the reason why I'm trying to do both of them.

If you have Usukane body and feet pieces and managed to get your hands on 5 more Store TP from Goading belt or something else, then you could 5-hit the OAT weapon and it would become much stronger than all the other options.

I realise this formula isn't mechanical enough to be completely accurate. I honestly don't know how to factor in Zanshin or if I should at all. I chose not to, because that's what makes the most sense to me right now. Another unknown is how Double Attack acts during weaponskills (some claim it can only proc on the first two hits). Having narrowed down my choice to two weapons and generally knowing which is better in which situation, I'm not bothered with being more precise than this. Still, if you see any mistakes or errors in logic, please let me know.

Monday, August 2, 2010

Very New Monsters (to me)

When VNM's were first introduced I was not interested. When they figured out strategies and all drops were discovered, I did not care. When I first heard drops from it were needed for OAT weapons, I was annoyed. The VNM system did and does not appeal to me as a hierarchical NM system at all. However, recently I was asked to help someone with Urd. I realised as little as I care for the system or its rewards, the fights themselves could be fun. So last week WoA popped Orcus.

All accounts recommended turtle/blood tanks as attack speed and frequency made shadows useless. Immune to slow elegy and paralyze it's pretty clear why there's little alternative. So, I read about its TP abilities. One is a hate reset. However, it's stunnable. I made the mistake of relying on the latter and brought only a single tank. Argonar has a pretty nice blood tanking set with PDT and defense. Muze was there on SCH to Accession Stoneskin and Phalanx. I could have subbed SCH and done the same, but Muze was able to put up Aquaveil as well and have time left over for more. We brought it to a cliff to avoid aoe effects through height differences. It worked but wasn't really necessary. Hate reset ability wasn't very common. None of the "special" abilities of Tier3 VNM's that require special measures are very common. Perhaps HP% dependant, but because of the lack of a visible HP bar I can't tell.

Anyway, having only one tank and Orcus becoming immune to Stun as we reached 20% meant Argonar couldn't recover hate at the end and as much as we tried stretching it, a wipe and depop was unavoidable. I did not expect TP abilities to be so frequent. I thought if we stunned each charge animation we'd get all the unique abilities. For the next time limiting stuns to a single stunner on only the big unique abilities should make it much simpler. Also, a second tank will give us room for error. I believe Orcus can be the easiest of all tier3 VNM's if we can get tanks spec'd the way we did on this attempt.

Today we popped the big slime. Fought it against a wall to avoid knockback. The first 3 Cytokineses weren't handled very well. I was late to pull them and BLM's were slow to nuke and at some point they were pulled on top of a resting Rcake who then got hate and died. BLM's also bit the dust in the first few leading to double weakness and some nerve wracking minutes. After recovering everyone focused more and it went much smoother. Towards the end we started blowing through it trying to cut the fight as short as possible. Dissolve lengthens the fight considerably. Its high resistance to Slow meant I took BLM sub to land it. A second ES slow woudl have helped, but it didn't occur to me to ask one of the BLM's to go for it. We were missing some people to Erase the slow from Mucus Spread. It went down and I think future runs would benefit from someone who can deal with the erasing exclusively.

If we handle adds like we did towards the end, it shouldn't be a problem. However, the tanking could be made much more bearable if we had one mage dedicated only to erase>haste and refreshing tanks. I tried to debuff NM, pull babies and buff tanks and backup heal, but it became too chaotic so if we split those tasks among two people it should be much better.

Lord Ruthven
Supposedly easy (and looking back I can see how) this fight we lost as well. Firstly, I won't ever pop it in Beaucedine again. Walking around that place takes too long. I asked Muze to come as DRK/NIN. It didn't work out very well as she complained about not being able to keep hate. Dirty Breath was taking a lot of Rcake's time and it draining buffs from tanks was slowing us down as well. However, DRK/NIN should be able to contribute very well and trump a PLD in enmity so I'm not sure why it wasn't balanced. Probably a combination of experience and macros with an enmity gear setup. Damage was slow coming from two SMN's and later a RNG. It would probably have gone down before rage from the pace alone, however it wasn't fast enough to avoid the enmity cap to be hit by Rcake. Nightly on RNG got hate several times as well, but that was no concern as it wasn't for very long stretches and he took no serious damage until the tanks went down.

Lessons to be learned were more Finale spam (Dispel heavily resisted), dedicated backup healer (as WHM will eventually start running low on MP) and more damage. ES Slow would be nice too. Wiki didn't mention its highly resistant and also claimed that 'Super Curse' was removable by Cursna, which given the "no effect" instead of "fails to take effect" message is very doubtful.

1/3 on Tier3 VNM's for me. Not happy about that record, obviously. However, it was my first time fighting any of them. So, I have an excuse! On numbers we were OK. They're definitely all downable comfortably with about 10 people. Even lower if you're familiar with the fights and focused enough. Now I know more about the fights than before and I suppose it's only natural that these things start out like this. All in the process. Still haven't seen Krabkatoa, Verthandi or Dawon yet. Ruthven really made me want to finish DRK so I can try tanking on it. A disengaged tank works well on Yilgeban too, but I have an idea how we can work that out.

Weekend update 7

Three summary posts in a row. I'll post something with a bit more depth soon. In the meantime, however, Magian trials!

A bunch of pickup groups got me through Kuftal Sabotenders and Lizards. Now, I thought, comes the hard part of killing 250 Rafflesias. Luckily for me, this trial is the same for a lot of the PDT weapons. There were plenty of people to team up with both times. That's right, it only took 2 earthdays. The first one got me 145 kills. A DRG was there to Sonic Thrust the crap out of the train the rest of us gathered up. It looked a bit like this.

That Weaponskill is perfect for this situation. The second group I got went at a slightly slower pace, but more than fast enough to finish the trial. We also killed the Rafflesia NM that saw me tanking all the charmed group members with my PDT gear. That was a lot of fun. So, PDT sword achieved! Maybe I'll go melee solo Lil Apkallu to try it out.

I also started on Great Katana trials. I'm doing both the Store TP Keitonotachi and the STR+Attack Efunotachi. Why two? Because I honestly think for most of my playstyle they are very close in damage and I want to see if my calculations match up to the real world.

In a previous post I said I was going for the Kasha+10% damage Keito. However, I made a little spreadsheet to come up with the overal DoT and that weapon lost to Hagun in most realistic situations where I wouldn't use Tomoe. I'll post the sheet and some more thoughts on it in the near future.

Dynamis - Tavnazia
Yesterday we had the best Tavnazia run to date. We were an alliance of 15 people. We dropped the NM's without any trouble and got to farming. Ground level mobs dropped very little, but Xarca mobs dropped a bunch. I got myself a Duelist's Tabard because Helly was kind enough to pass on it. Thank Helly! Duelist's Chapeau still eludes me.