Related it would be good if the nameplate length limit was lifted.
As it is now if we are using color codes we must cut down the name to 16 characters to avoid breakage. For the longest usernames that results in cut down names. This also applies to the playerlist(think it is related).
I've developed a proxy that if nothing else passes on the messages as is. it has been working for over a year so I am rather sure the test changes I do to trigger these bugs are bugs in the client.
The bug is about the client crashing, many times one can send bad packets to a vanilla client that makes it disconnect from the server with an error message, that is OK.
However the vanilla client should never crash caused by correctly structured data but unknown values.
Also since the extra 4 bits were added to the map format that will in the future allow for a lot more blocks than build into the vanilla client I assume it must be able to handle unknown types.
I have only done testing towards the client, but the nature of the bugs makes indicates that similar bugs could be found server side. So finding the core issue here would probable help.
@Ray Well our proxy is only causing it when I tell it to such as when I want to generate the crash log for this report.
Yes the Minecraft protocol does change and we catch up all the time, but the development would have been a lot easier if the vanilla client simply disconnected rather than crashed and held the computer for half a minute.
I did send a similar report for another but that was closed since it did not affect vanilla client <-> vanilla server. I'm still stand that ones client should not crash for packets it receive over the internet, hover since you are of a different opinion you should close this issue too.
At least it makes a good tool for putting force into the /ban command
Nuxas
MC-106![[Mojang] Nathan Adams](../../../data/assets/gravatar/dinnerbone.png){kind=small}
Related it would be good if the nameplate length limit was lifted.
As it is now if we are using color codes we must cut down the name to 16 characters to avoid breakage. For the longest usernames that results in cut down names. This also applies to the playerlist(think it is related).
I've developed a proxy that if nothing else passes on the messages as is. it has been working for over a year so I am rather sure the test changes I do to trigger these bugs are bugs in the client.
The bug is about the client crashing, many times one can send bad packets to a vanilla client that makes it disconnect from the server with an error message, that is OK.
However the vanilla client should never crash caused by correctly structured data but unknown values.
Also since the extra 4 bits were added to the map format that will in the future allow for a lot more blocks than build into the vanilla client I assume it must be able to handle unknown types.
I have only done testing towards the client, but the nature of the bugs makes indicates that similar bugs could be found server side. So finding the core issue here would probable help.
@Ray Well our proxy is only causing it when I tell it to such as when I want to generate the crash log for this report.
Yes the Minecraft protocol does change and we catch up all the time, but the development would have been a lot easier if the vanilla client simply disconnected rather than crashed and held the computer for half a minute.
I did send a similar report for another but that was closed since it did not affect vanilla client <-> vanilla server. I'm still stand that ones client should not crash for packets it receive over the internet, hover since you are of a different opinion you should close this issue too.
At least it makes a good tool for putting force into the /ban command
@Tails if you can confirm this please change confirmation status
Some suggested a new gamerule as solution for this issue, a new gamemode would be much better.
The important difference is that gamemodes can be set per player and thus allow builders and players whereas a gamerule would have limited usefulness.
How about showing a warning dialog that informs the user that there is leading space in the address and tell them to remove it</irony>
also /kill from the client only works for admins.
Please update the version, this still affects 1.5
This is probably not a duplicate of
MC-3766.That ticket report a client crash, in this issue the server and clients are still running but reporting the error log as written before.
The exceptions are different and at a glance I can't see any resemblance of the stack traces either.
A workaround is to block the AttachEntity and EntityMetadata