[IC2 1.109] SeedManager v3.0.2 - Taking the hassle out of agriculture.

Quote from Kane Hart

And rewrite this mod from scratch

Nonsense, I'm not rewriting my mod from scratch. I'm just rewriting MCP from scratch.

(And I'm only half kidding about that.)

The good news as far as SeedManager 3.0 is concerned is that I've acquired some personal matters (read: job hunt) that I need to attend to weekly, and working on this is my major justifiable procrastination.

Basically, what I'm working on is making MCP stop being stupid. Any time you want to compile a mod with MCP, you have to recompile *all* of your mods, plus Minecraft itself. And then reobfuscate all of the above. Meaning that it takes around a minute each time you want to see your change in action. This makes debugging... painful.

One day of coding already gave me a working replacement for RetroGuard (the obfuscation/deobfuscation program that MCP uses), albeit with one annoying limitation: It doesn't understand inheritance. Another couple hours got me the framework for learning the inheritance, so my next project is linking them together and making it smart enough to track inheritance for classes that aren't in the current batch, probably meaning some sort of persistent storage.

Once that's done, the biggest obstacle to optimizing MCP is gone, because I'll be able to reobfuscate a mod by itself, without processing all of Minecraft. From there, I'll be integrating mcp_rebuild and improving it to do single-mod compilation. Run a single-mod compile, package up a zip file, reobfuscate it, and presto!, you've accomplished in a couple seconds (or less) what took a full minute before.

Better yet, I'll be able to run mods through a "deobfuscation" to the MCPC Bukkit naming, which will make it possible to build a Bukkit version at the same time. It's not magic, though; I'll still need to make a couple adjustments and maybe even a small code fork to account for the Bukkit idiosyncrasies. But the distinction won't be any worse than the previous client/server split.

So, yeah, rebuilding MCP from scratch, sort of. But it's not that bad. The hardest part (for me, at least) is figuring out how to tie stuff together in a way that makes sense to the user, and even that's not too hard. The really tricky part of MCP is building the renaming tables, and that I can just use as-is.

All-in-all, I'm guessing something like 5-7 days of work. Could be less, but I'm too familiar with Murphy to expect that. As long as I can stay focused (fingers crossed!), I'll have something polished up by the start of November. (Again, could be faster, but Murphy's a dick.)

Good to know. I'll be sure to stop by the forge IRC and prod cpw about it when I get to that point.

Quote from mcmish

hey , is there a version of this compatible with tekkit server 3? I've got the clients running, but the server just gives a lot of severe code when i tried the bukkit ported version on page 2 of this thread. Amazing mod though

I don't support Tekkit. Unlike some, I don't mind Technic, but I've no desire to support a random conglomeration of three bazillion mods.

Quote from NLZ

Could you get anything out of cpw?

That comes soon. I've got a combination of mcp_decompile and mcp_rebuild that seems to be working properly. It successfully processed SmoothWater from source to working Minecraft, but everything else I have requires Forge, and most of it needs IC2, too. So non-trivial testing needs more groundwork.

My next projectlet is a script for deobfuscating and decompiling mods. I only really need the first half, but Searge says the decompiling would be very well received (and I think he's right), so I might as well tackle that, too. After that, there will be a little wrinkle with allowing mods to build against deobfuscated mods, and then I'm on to Forge and integration tests, which is about when I'll hop into the Forge IRC room.

Quote from "Hewlitt"

I'd love to use this in my modpack, do I have your permission?

If you're making a modpack for 1.2.5, go right ahead. You're welcome to *try* on 1.3.2, too, but you may find it ever-so-slightly difficult.

For the curious, you can watch my project on github. master is the main part (pretty much done), and mcp_interface has the user-level stuff (plus an entire MCP install, for proper diffing).

I've effectively finished everything pre-forge (the mod decompile was so trivial I haven't written it down yet; literally just unzip -> fernflower), so I prodded cpw a little. From what he said, I gather that his project is essentially orthogonal to mine. Same basic goal, different approach, and since he's deprecating his project that I would have been building off of... there's not really anything for me to do on that front.

I'm really close to building a new version of SeedManager. It'll be a test case for handling Forge, which is next up. So a new version as soon as tomorrow is possible.

Well, I'm utterly unsurprised that it took a few more days, but I've got a 3.0 pre-release ready.

Important notes:

• Seed analyzer blocks in any inventory (yours, chests, etc.) will magically become Seed Library blocks when loading an old save. This is unavoidable, due to some internal restructuring. Placed analyzers should be unharmed, though they may look like libraries until you give them something to do.
• Both blocks only have one orientation for now. This will be fixed later; the difficulty is in showing the orientation of the top face.
• SeedManager is now using a universal package, and seems to be working well in all setups.
• The save structure changed slightly in 3.0_pre3, to support filters that only match unknown type/GGR seeds. Older saves will update automatically, but if you downgrade to an earlier version, the missing info settings will be lost.
• The GUI has received some love, and more is incoming.

Did you read the notes? I'm not kidding when I say that they're important! Yes, I know it's buggy.No, I just want the new version.

I've updated the pre-release download link. It seems to be running properly on true servers now, and I fixed a display failure caused by being in negative coordinates. (I hate java's broken % operator.)

Version numbers, please. Forge especially; my guess is that you're on a non-recommended build (like the IC2 1.107 beta requires).