Work blog mentioned on the stream

Lots to read in it, but I wanted to share my up to date work blog. Most of the items are behind the scene elements but lots of player facing items as well.

https://catnipgames.net/

Screen shots to come as well.

Discussion thread: https://www.shroudoftheavatar.com/forum/index.php?threads/atos-workblog-discussion.181916/

 

As promised during the stream, I expanded the localization to handle multibyte, kanji, mandarin, simplied Chinese character set (only 3500 characters, simple!) Plus, all text in game converted.

 

Also, screen shot of new patcher. Uses CDN for initial download, then rsync like tech over standard http to be more office friendly than the old Torrent version. Dramatically simplifies our efforts as well when it is patch time. Also, you can choose graphics engine as well as language from the tool.

Unlike the previous patcher, this version is done in a native app that is a single, 45mb file. Rebuilding it takes seconds and it self patches itself automagically!

 

As I work through bugs and see there are now effectively 3 types of COTOs in data. One for the old coto, one for the new coto, and a third for the coto that was purchased vs dropped. My memory on this is that it was done because some people were screaming loudly that we sell cotos for money and therefor if we let people sell them for gold, it would be pay to win because you could convert $$$ to gold.

The reality is the result is we've complicated systems, confused inventories with two nearly identical cotos that don't stack, and no one in their right mind would EVER buy crowns to sell them anyways because it is 1000x worse than most other ways to get gold or what you would get selling to another player.

We're going to simplify this and there will no longer be two different types of stacks of COTO. Instead we'll remove the gold value from all of them, dropped and purchased, combine them all into a single type. If you want to sell them to a merchant for gold, first, you're bad at math, and second, do it before October, when we hope to move to the new rule set!

tl;dr single COTO type with no gold value starting in October

 

Today's fixes going into the next "Chaos Shard" test server build:

- SOTA-49862: Crafting stations now have an "Only craftable" checkbox. Recipes you are missing ingredients for are dimmed and sorted below the ones you can make now, and the setting is remembered between sessions.
- SOTA-72490: Added an "Auto-Decline Party Invites" option to the Multiplayer options, matching the existing auto-decline for duels.
- SOTA-72122: Gold Crowns of the Obsidians now stack no matter where they came from, so a major repair works when your crowns are split across purchased and earned stacks. Existing split stacks can be combined.
- SOTA-65504: Location sharing is now a three-way choice: Everyone, Guild only, or No one. Guild only shows your location to guildmates in the social list without broadcasting it to everyone.
- SOTA-72042: With "Prevent Movement While Casting" enabled, pressing Jump mid-cast no longer cancels the spell. The key is ignored until the cast finishes.
- SOTA-72190: Friends now show in green in the Players in Current Scene list, so you can see at a glance who you have not friended yet.
- SOTA-72607: Ctrl-U highlights planter boxes that have an open planting spot, the same way Ctrl-W shows unwatered plants.
- Vendor buy-order search: the Vendor Listings window has a new Buy Orders toggle. Search who is buying an item across all zones, sorted by the best offer per unit. This one is already live on the test shard as of build 23.
- Fishing XP now scales with the catch's trophy length instead of a flat per-zone amount.

Except where noted, these reach the test shard with the next client build.

 

Yesterday evening and probably most of today will look slow as I'm working on a BIG task, replacing the core networking library we use, Photon. This is a critical next step for a few reasons. First, it is the only piece of non-open-source software we use on the server. While we have a permanent license for Photon, if we ever wanted to spin up additional servers or release the server to players, replacing Photon would make that much easier.

Reason number two is Photon server is the last piece of Windows software left in our tech stack. If we can replace it, then our entire stack can be Linux, top to bottom. Previously, our tech stack for the server was Windows across the board. One of the first things I tackled upon setting up a fresh shard was moving everything to Linux BUT Photon. The current new shard is Linux for everything and then a single headless windows VM running on the linux stack for the Photon servers.

In total, this means replacing almost 650 calls to Photon, and that is after building a new library that can do roughly the same job as Photon, which in itself is a big task. So far, it is moving way faster than expected. Currently still in the build-new-network-library side of things, but already getting close to having that fully functioning. For the nerds in the audience, step 1 was just identifying all of its responsibilities and documenting them. Then identifying where we fit the server tech to the library vs what we actually needed to see if that helped simplify things at all. The third step is building a system to fully replace our new, smaller requirements. On the fourth step now, with everything fully documented and itemized, I'm now building out a mountain of unit tests to prove out the tech across the board and also to benchmark things as we go.

The good news is that we don't have to match Photon's performance. A single Photon server can handle thousands of connections. Because we cluster our game servers, we really just need to be able to handle a tenth that many.

Just wanted to share for the nerds in the audience and also so people didn't think I was losing steam just because I had fewer items in my work blog today. It really should just read: "Building and testing a new networking library" and if that is all it said for the next week, that would still be a HUGE lift!

This isn't a make-or-break item, but it will definitely make everyone's lives easier going forward if I can pull it off! Wish me luck slaying the dragon!