Announcement 1.13, debug symbols and the Titanic

Recently there's been a lot of drama going around about the removal of debugging symbols from a recent 1.13 beta.

So what are debugging symbols?
For the non-technical user, consider these as an essential tool on which every bit of server and mod development has relied on for the last 7 years.

For the more technical user, debugging symbols are names of artifacts located within an executable or library. Such things enable us to find the names of functions, constants and more within the Minecraft binary. These debugging symbols are not necessary for a binary to be executable, but they are critically useful for reverse engineers.

Why is it such a big problem that these debugging symbols have been removed?
The removal of these symbols is a very big problem for server development, because reverse engineering the protocol involves taking apart the Minecraft binary in reverse engineering tools such as IDA, and extracting information about packet structures and functions.
When debugging symbols are removed, it has the same effect as obfuscation, and makes it very difficult to extract meaningful information. Automated tools of the likes created privately over the last 3 years become useless, and extracting information from the binary by hand also becomes far more difficult.

The bottom line is that updating for each new Minecraft version is about to get exponentially more difficult and therefore take a lot longer. it is going to be significantly more demanding on my already very constrained resources.

What implications does this have for servers?
if anyone continues to maintain PocketMine-MP, updates are going to start taking much longer. Right now updates take a couple of days. Without symbols we are likely talking about weeks or months for a new version to be supported.

You may have already noticed that updates are taking longer since the 1.9 release. The reason for this is because I have stopped allocating my resources to updating for betas. Chasing betas used to give us a head start on each new version, but the time cost necessary to continue doing this became far to excessive, considering my lack of funding and limited available time.

What are you planning to do for a future version updates?
Chasing betas will become necessary in order to keep up with the existing state of affairs after symbols are gone. This means that to continue receiving timely updates I am going to need substantial and sustained extra resources from the community, or there will be significant delays in rolling updates.

I have also considered simply walking away from PM development and leaving it to rot. The return on investment I currently get is far insignificant to the cost of the time and effort that I put in. With this development, this problem is only going to get worse.

Closing notes
This development spells trouble for the third party server ecosystem. We are at a crossroads in the server ecosystem. It's up to the community to decide how they want to react to this.

 

Addressing some common questions

OMG! iS pOcKeTmINe gOiNG tO dIE?
I think it should be obvious that I wouldn't make so much noise about this if it was guaranteed to die. I could simply walk away without all this nonsense.
As stated in the original post, I consider quitting as an option if the work becomes too arduous or if it consumes too much of my already limited resources. If you want to alter this, consider contributing code or donating.

Can't you reverse engineer BDS (Bedrock Dedicated Server) instead? Doesn't that have debugging symbols?
Several people have suggested using BDS for reverse engineering instead of the Android client. While this suggestion has merit, there are a number of problems:

1. We have no idea if BDS will continue to have symbols when 1.13 is released. Since there are no BDS releases for betas, we don't know what the state of affairs is there yet.
2. BDS is a server-only executable. This means that it's not possible to use BDS to reverse-engineer client-only functionality, i.e. it only tells us half of what is going on. The Android binary is both a client and a server, which means it gives us both.
3. All of our tooling is designed around the Android client. There are some tools that will only work on Android (for example my invaluable Frida.re packet tracer)
4. As stated above, BDS is only released for stable MCPE versions. This means it would cease to be possible to get a head-start on new releases by reversing BDS.

If BDS is still suitable when 1.13 comes, and we still don't have a client with symbols, then it will likely become a necessary survival tool. However, the ability of server developers to implement the protocol will be impaired due to the reduced available information.

What about MiNET, NukkitX, and other server software?
Their developers are all dependent on the same information that we are to implement the protocol. No one server software is going to get this easy.

Can't you just copy paste from Steadfast2?
This suggestion has come up multiple times.
Observant developers may notice that PocketMine-MP has never relied on Steadfast2 to implement new updates in any way, for the following reasons:

1. It is an unholy mess, and it's almost as difficult to extract meaningful information from it as from MCPE itself.
2. It is an incomplete reference. The developers of Steadfast2 only implement what LBSG needs. Large parts of the protocol have not been implemented in Steadfast2 because LBSG doesn't need those parts.
3. Its protocol implementation is incorrect in a large number of places. Relying on it would compromise the quality and reliability of PocketMine-MP runtime (and also its status as protocol documentation).

I have used Steadfast2 as an additional reference point for updates in some cases (to reaffirm my own assumptions), but I have never directly relied on it.

How do you know that the removal of symbols is intentional?
I can't directly reveal my sources, but I have contacts within Mojang and within various partners who have all confirmed the same thing.

Why are Mojang trying to kill servers?
I've heard this come up multiple times, and it's simply untrue. The reality is that we are simply caught in the crossfire of a different battle. Mojang is (ostensibly) trying to make the lives of hackers more difficult. Third-party servers are just collateral damage. The fact that such a move will not stop hackers is typical Mojang irony.

 

The TL;DR

 

TL;DR: Mojang has split personality.

 

yup, except it turned out nobody (including me) fully appreciated the scope of what I want to do.