asiekierka

I tried making Minecraft-quality mods for Minetest and the API just wasn't there and... I stand somewhat corrected. Some of the features I wished were there are now there, but not all. So it is moving forward.

>The problem with minetest is nobody mods it so you can't program a turtle in LUA on minetest, only minecraft.

Minetest has quite a lot of mods, admittedly - I even made a few. The problem with Minetest is that it has too many modders and not enough players. The problem with Minetest is that the voxel engine zone itself is fragmented: Minetest itself has three different maintained versions, and there's other clones like Terasology as well. The problem with Minetest is that the APIs are and will probably remain too limited for the most complex things, seeing as it's not the kind of modding where you turn the engine inside out for fun and it happens to work with most mods for some reason - and, even then, there's far too few developers on Minetest's core engine and far too many opened/unmaintained pull requests.

The "Adventure Update" and future updates in the same vein: hunger, enchanting, RPG-esque elements, etc.

Also, some of the major redstone additions from the later versions, which are the ones I'd miss most.

It doesn't. Many classrooms rely on ComputerCraftEdu, for instance. While OpenComputers exists, it is not optimized for education, but rather for gameplay.

Actually, Pocket Edition does have a modding scene - they left all debug symbols in, which let a few willing developers build an API around it. Windows 10 Edition, however, does not, and is thus not moddable.

Pocket Edition is a separate codebase, used for Windows 10 Edition which is not yet however a replacement for Minecraft.

There are, I believe, three distinct versions of the engine: the Java one, the C++ one and I think the console codebase is also separate.

There has been severe refactors going on ever since 1.2 or so. Whether they are beneficial or not is still a large debate between modders, but from a stability and customizability perspective they most likely are.

I played with Minetest for a while. It feels like it suffers from a lack of identity beyond "it's Minecraft, but free and hackable!". This is not enough to build a unified playerbase, I'm afraid. Its API is much simpler, though, if limited (no client-side scripting, limited model options).

Sadly, TrueCraft, due to its nature, has to lock out one of the most potentially interested groups - Minecraft modders, both former and current. This saddens me a bit, as I myself am a strong fan of the beta 1.7.3 design ideals.

Orebfuscator is a pretty nice solution to X-ray. As for super fast mining tools, you can just not add any to the game.

Any attempt at improving the mod mechanism will make many of the things mods now do impossible, as many of their features rely on things which would never fly in a sandboxed environment. This is probably why, instead of the modding API, we get Mojang hiring the developers of the Minecraft Coder Pack.

The license seems to be "very illegal", as this is a fork of the NT4 source code leak.

NetSurf is not keystroke-centric enough, obviously.

(Also, uuterm is much lighter if you want a framebuffer-based terminal, however it needs a patch ( https://github.com/asiekierka/yuyulinux/blob/master/uuterm/u... ) to support 32-bit FBs properly.)

Yet. The Git version seems to already have preliminary JavaScript integration, based on Duktape. There was also an earlier attempt in 2012-13, utilizing SpiderMonkey.

Have you looked into Polish tax laws?

Spigot does this for another reason: thanks to someone making the arguably stupid choice of using GPLv3 for a project linking to (propiertary) Minecraft source code, CraftBukkit has been DMCA'd by one of the lead developers over Mojang taking over the project when the lead dev team quit and is now essentially illegal to redistribute.

Try using RedLogic. It's what Project: Red forked off of and provides better performance at the cost of, well, only having RP2's logic components.

(Or just update to 1.7. Most mods should update seamlessly.)

This looks like a great proof of concept with some software (especially a lightweight package manager) which I always wanted to see come to life and possibly use myself!

From my experience, this bug appears rather randomly.

Heh. Nice work.

As a mod developer, my plan on revitalizing Minecraft and making it more of a sandbox again was always centered about designing a modpack going in that direction - I've been working on it since December.

I don't really feel splitting the community into twenty reimplementations is a good idea myself, though.

Freeminer is trying to improve on Minetest in the graphical aspect, if you're curious.

>I hope you make it so that only the surface blocks are sent to the client. The time it takes to "mine" a block is enough for the latency to send the block behind it.

Orebfuscator or Spigot.

>Also, send the input to the server, and not the position/speed vectors. That way you will also get rid of the speed hacks.

What? No. You can't make sure everyone moves the same way, and this would undoubtedly lead to desyncs. If you want to make sure players don't move too fast, use NoCheatPlus or what it was called.

A far bigger issue for servers is mods like Gammabright which make all blocks have a brightness of 15, thus not requiring you to use torches or light sources, as that's very hard to work around.

Garbage collection pauses work very much fine on a heavily modded server I monitor as part of BuildCraft development. The server, with its 40+ mods and about 20 thousand tile entities running at any given time generates about 50-100MB of objects every second. In fact, Minecraft's own explicit chunk GC is MUCH slower than Java's and causes actual 2-3 second pauses!

>Java was definitely one of the reasons the original Minecraft was bad though. It (along with bad architecture) was the reason behind the massive server footprint and why hosting minecraft servers is extremely expensive and completely unscalable.

No. It wasn't Java.

The reasons are many, but here's some: All world logic happens in a single thread (fixed in Minecraft 1.8, now there's a thread per world). Entity lookups are very badly coded (mitigated, but not fixed by CoFHCore which explicitly excludes dropped items from the lookups). There are many bad coding decisions made throughout Minecraft (a nice amount of them is fixed by FastCraft, which optimizes Minecraft and reduces its RAM usage by ~300MB; OptiFine used to be good for the GPU but 1.7/1.8 integrated most of its fixes). Every neighboring block/tile entity access loads that neighboring chunk, then marks it as unused, then unloads it every 30 seconds (not a huge issue in MC itself unless you use a lot of pistons, but a massive issue for modded servers)

Etc, etc, etc. It is possible to scale Minecraft, however nowadays it is not possible to scale Minecraft without mods and at least a basic knowledge of how the engine works.

There was an experimental mod made by nallar in 2013 which dynamically ASM-patched Minecraft and every mod to make all world logic fully threaded in multiple worker threads, as well as a separate thread for chat; however, that was really hard to maintain in practice, especially as mods grew more and more complex.

A commonly used method is to run multiple instances of Minecraft servers and use a proxy glue server like BungeeCord to treat them as separate world on one server.

The Plan 9 support was contributed by other people. I'm sure that if there was nobody left to maintain it, it would have been dropped long ago.

I analyzed this.

"t" is a Base64-encoded blob of data containing a few miscellanous bytes, as well as the URL the proxy should download (in this case, twitpic.com/show/large/...). "s" is the signature, used so we can't just abuse the proxy to download more.

The Archive Team only backs things up when there is a risk of them being gone.

They cannot do it any other way - the Internet Archive does not have infinite resources to host things, and they need to prioritize. 120TB is not an easy thing to host.

It's only downloading comments and metadata, not the images.

You must not forget that many distros moved to systemd out of necessity rather than desire, and for some (Debian) the decision was lengthy, splitting and controversial.

EDIT: By "necessity", I mean things such as GNOME (which is a very often used graphical shell in a majority of distributions) and udev (which is getting integrated into systemd, is used by most distros, and, bar Busybox's mdev or Gentoo's eudev fork, has no alternative) depending or planning to depend on systemd.

He does.