Drakkin Features

[KLS]

I would suggest against that, what if we later find out what the field we stored it in means? You then have to write a utility to convert every character over. I would suggest the extended player profile.

[trevius]

(KLS, I know you fully understand what I am explaining below, but I am just trying to explain my understanding of it so others can follow and so I can be corrected if I am wrong about something. And I am trying to figure out if we should really care about the blob struct anymore.)

I figured that the profile blob didn't really matter at all. It is essentially used as a table inside a table. In that we pull out the blob, pull the data from whatever piece of it we are needing and then save changes if needed and put it back in. Seems like we could arrange it in any order we wanted. It doesn't send the actual blob itself, unmodified. It creates the profile 1 field at a time by us telling it where to look in that blob based on the structure we have set for it.
There is so much wasted space in our current profile blobs, that I am sure it eats up a fair amount of space in the DB. Any unknowns in there are just a waste as far as I can tell.

One problem I have is that all of our clients have encodes set on the player profile. This means that the structures for our clients do not match our blob structure. So, even if we did ever identify new fields on either Titanium or SoF, the player profile blob is completely useless to us for that since it most likely isn't going to line up perfectly anyway (they are very different). So, we have a ton of fat on the current PP blob that is just there for no reason as far as I can tell. We should probably either try to use it, or get rid of it.

Just a little math (and yeah, sorry I am getting off subject lol):
The player profile structure for the blob is currently about 20,000 bytes (20k)
If PEQ has 100,000 characters (just making up a number), then their total space used just for player profile blobs is about 2GB (100k X 20K)
That makes for a fairly large database backup file I am sure. So, over half of that blob is unknown, and I believe completely unused. So, they could be saving 1GB of space if the unknowns were removed. The Storm Haven full database backup is about 2GBs and 500MBs of that is just the character_ table alone, and I am sure almost all of that table is player profile blob. Not that size is really an issue, but just giving an example of how much space we are wasting with the unknowns.

Since we do encodes of the whole thing anyway for all of the expansions we use on EQEmu, couldn't we just make up our own PP structure to however we wanted it? I am sure it would be a hassle to convert it over, but other than that, I don't see why not. I personally would love to see it moved into it's own table, but that would probably be a nightmare to write that all up.

Sorry again about getting off of the subject, but I just think the old player profile is not as useful as it probably once was. Before needing to do the encodes, we could probably just send that whole thing in 1 big chunk! But now, we break it down and put it back together again however we want it, so it probably isn't that great of a way to handle it anymore.

[KLS]

Alright but you gotta make sure you write the encodes so they work. And remember once you put something into the pp it's kinda hard to take it out~

[trevius]

Still trying to figure out exactly how I want to start implementing these. Maybe I will try to use the extended profile after-all, just to get away from using the player profile blob. Maybe some day we can do an overhaul on the PP storage or replace it with something better and easier to manage.

Seems like it is going to be quite a pain to get these new fields added. Going to need to update the npc_types table with the new fields, and update the extended profile with the new fields. Then, going to have to update all of the related structures and encodes/decodes where appropriate. I think that will include the char select, char create, spawn, player profile and facechange structures at the very least. The hardest part is probably going to be making sure that the way these structures get that data are all setup properly. I haven't added new fields like this before, so this should definitely be a learning experience. Hoping not to mess anything up in the process, but I will try to test as much as I can to be sure before updating the SVN with anything.

Before doing anything with the new features, I need to make sure that Titanium facial features are all working properly without the previous hacks. I already removed the hacks and SoF is working perfectly as far as I can tell, but now I need to get Titanium working 100%. Once that is all done, I will try to get started with adding the new drakkin features. It's gonna be cool to have this stuff finalized

[realityincarnate]

Sorry to be the bearer of bad news, but it looks like at least some of the Titanium features are wrong now. I haven't looked into it in depth, but one of my test characters is now bald (no idea what he used to look like, I hadn't looked at him for a while, but I know he had hair), and Arias in gloomingdeep has a beard and combination ponytail/baldness. I know he's never looked right, but he never looked like this either. Those are the only two issues I've seen so far, but I haven't looked much either.

[trevius]

Yeah, I am aware that Titanium isn't working perfectly yet. I think it is a structure issue with the spawn struct and maybe the player profile as well. SoF seems to be working perfectly from the limited testing I did so far. The only reason I updated the SVN with this change before it was complete is because there is quite a bit of updating that had to be done to clean some stuff up. The actual change that is affecting Titanium would be easy to back out of if needed, but I am pretty sure I can get it working perfectly and properly as well.

Also, even after the changes go in to correct how this stuff works, people will almost certainly have to do a facechange in game to get their character looking the way that it used to again. I am sure that would be the case if it is actually a structure issue. Though, if it is some other non-structure related issue, then they may not need to. I think as long as NPCs look correct after the change, then the change should be a good fix. I know most NPC settings were pulled directly from EQLive, so for those, they should mimic live appearance exactly. If they don't, then we have something wrong with a structure (which is my best guess to the problem atm).

I think one of the things that might make this change more complex is that it is hard to know which NPCs in the database are exact copies from Live and which were added manually by EQEmu people. Also, the hack that was previously in place may cause some unwanted effects after this code is corrected. Basically, with players, any features that were set to 0 were being converted to 99 and saved and then when they were loaded again, it converted them back from 99 to 0 again. So, since after my change, it no longer converts 99 back to 0 (or the other way around either), anyone who had a feature that should have been set to 0 would now load 99 instead, which is always bald or broken features. This may be the same case for NPCs as well, because it seems like those were all being converted to 255 (0xFF) and saved and then converted from 255 back to 0 when loading them again. There may be a simple solution for NPCs to correct this though. We could probably just run a simple query for each of the feature fields in the npc_types table and convert anything that is set to 255 down to being set to 0 as it should. I haven't really looked too closely at the table to see what is going on with that yet, but I will.

I am pretty confident that we should be able to get this working 100% perfectly for any client without having to use the previous hacks. The only thing that we may need to do special is that we might need to put certain conditions for some of the player races so that they automatically block certain fields from being set. I haven't fully tested that just yet, but I will try to tonight if I have time.

***EDIT***
After looking over my own DB (1.5 year old PEQ included), it seems like most of the feature fields look normal and none of them are set to 255 with the exception of the face field. For some reason, it seems that about have of them have face set to 255. It is possible that this is due to an issue with the code that was previously doing those conversions, but I am not quite sure.

[Shendare]

I can confirm that player characters in both SoF and Titanium are showing bald (and probably incorrect beards, colors, etc.) since the in-game fix to SoF. This appears to be because of the old code that used to change zero values for appearance fields into 99. The characters still have 99 recorded in the fields in the database.

Uncommenting this block of code in worlddb.cpp fixes the problem:

worlddb.cpp - Line 90

  //*
  if (pp->face == 99)       {cs->face[char_num] = 0;}
  if (pp->eyecolor1 == 99)  {cs->eyecolor1[char_num] = 0;}
  if (pp->eyecolor2 == 99)  {cs->eyecolor2[char_num] = 0;}
  if (pp->hairstyle == 99)  {cs->hairstyle[char_num] = 0;}
  if (pp->haircolor == 99)  {cs->haircolor[char_num] = 0;}
  if (pp->beard == 99)      {cs->beard[char_num] = 0;}
  if (pp->beardcolor == 99) {cs->beardcolor[char_num] = 0;}
  //*/

That will reverse the old 0 -> 99 code. Note that cs->hair[char_num] in the fourth line must be fixed to cs->hairstyle[char_num] for it to compile.

Alternatively, logging the bald character in and using the Face change feature to reset the affected appearance fields permanently fixes the problem.

I couldn't get beard colors to work in Titanium, but beard styles were fine. It's probably a matter of figuring out the correct field location for Titanium beard color in Spawn_Struct and CharSelect.

Also, I didn't see a problem with Arias in Gloomingdeep with Titanium. On my screen, he's a normal brown-haired male human with a moustache, wearing leather armor. He looks the same to me with the SoF client.