The middlewere is able to distinguish between an existing and a new connection.
Given that the middleware is able correctly direct the byte[] to the right LogicCore, we could remove the separated management of incoming connections and pass those through the

below are accepted comments on which code style we should embrace.

My first suggestion are below:

• interfaces start with an I ( like IPCPpacket )
• every base class must first have a generic interface

This issue will track the state of the SPMP document

• iteration plan
• release plan

This is a reminder

This issue will track the state of the SRS document

version
PCP-Min.b.4

Describe the bug
The interpret doesn't use the chosen charset (ISO/IEC-8859-1)

should we maybe refractor the code to allow future expansions of the protocol?
maybe by moving all packets to a Min package?

reference pull request is #23

We have to decide what port to assign to the server.
I propose <<53 101>>: "5e" in ASCII

version
PCP-Min.b.4

Describe the bug
Null pointer exceptions on disconnection

i mean packets with extremely simple logic like the error packet shouldn't require a custom test.

version
PCP-Min.b.3

Describe the bug
A client that registers for the first time does not receive the list of users.

Expected behavior
A client that registers should receive both ack and user list.

this is a reminder to write documentation for everything.
Doesn't have to be done right now, but before the project finishes, this should be done.

class structure is sill to define, but should be easy.
1 class for every major package type, all packages are in the PCP.packets java package

version
PCP-Min.a.2

Describe the bug
Incomplete data packages time-to-live is not updated before the cache cleaning death reaper deletes them even if still valid.
Set accessing methods should be exposed from PCPManager.

Expected behavior
datas time-to-live should be updated when the interpreter interprets new data

ok so you guys need to learn how junit tests work.
This is kinda an "officialization" and a reminder.

Meanwhile you could look at some documentation and videos online.

simple reminder to test every class in the APIs.
The important thing is tests are well written, but the code coverage goal is at least 40%

I've looked through all of the packet classes... I think some properties are of the wrong type.
especially in @Alessio789 's classes.

• i think the assigned Id could be considered as an int or as a byte[], since it's a numeral, I don't see why it should be of type string. That's thoughout all of the classes.
• json content in the PCPgroupUserList class can be a Collection or an ArrayList and operate the json string conversion while converting the package, or even creating a method, like jsonStringContent()

• Registration

• Disconnection

• Alias change

• Group users list RRQ

• Message user to group

• Message user to user

• error

1. As implemented with commit 18642f5 classes now must implement byte[] header( ).
   I suggest using this method to simplify Collection<byte[]> toBytes() implementations.

2. documentation for every class must be provided (obviusly not for overriden methods, as java overrides the documentation too).
   We need to close this first -sprint- and move on to develop the APIs asap. Clock's ticking, gentlemans.

need to design the structure of the server and the libraries

version
PCP-Min.b.5

Describe the bug
In PCPMinInterpreter the accepted length for the topic ranges from 6 to 32 byte instead of 3 to 64 byte as specified in PCP-Min.md

version
PCP-Min.a.2

Describe the bug
calling PCPServer.shutdown() doesn't finalize all of the resources.
Probably some threads are still running unmanaged.

To Reproduce
Steps to reproduce the behavior:

1. initialize a server
2. try to stop a server in code
3. the program will still be running

Expected behavior
The server should shut down and release all of the managed resources and interrupt all threads

Additional context
Tested on an Ubuntu Linux 18.04 machine.

Proposed solution
A disposing mechanism should be implemented for all classes in the server structure.
something like an IDisposable to allow deterministic and safe resource cleanup

the scheme works perfectly if we think about already established connections...

But i think we need a specialized logicCore with the sole purpose of establishing new connections, meaning it must implement methods with a reference to the original socket.

Like, instead of a queue it should contain two: one referencing sockets and the other the data recieved.

This way it could be able to update PCPSockets with the login information.

Feedback?

Because my idea if the refractor suceeds was that we might delete this release and do a new one and call it Min.a.1b

a simple client. initially for testing purposes.

this issue will track the writing of the vision statement document

version
PCP-Min.b.3

Describe the bug
The method is executed even if the user list is empty.

Expected behavior
The method getAliasesByRoom() must control the number of users before proceeding.

looking at it I've begun to think that message classes need to inherit from a major abstract APCPMessagePacket class (A is abstract).
The structure would be as following

  IPCPpacket
      ↑
AbsMessagePacket
↑
|- UserToUser
|- UserToGroup
|- Recieved

version
PCP-Min.b.5

Describe the bug
Time to live is not updated when new data is sent from a channel

Being the doubtful insecure idiot i am, looking at it again
got me the doubt that maybe this code design is a bit overkill or even plain wrong.
Initially i thought it might be a nice level of abstraction.

Opinions? I would like to know what @antosette has to say about that.

We have to checkout if the PCPInterpreter is well implemented

version
PCP-Min.b.3

Describe the bug
the garbage collector is too aggressive.

Implementing IPCPInterpreter

edit documentation with the following:

protocol

• topic managment
• only one public room ( general )
• utf-8 charset (ASCII)
• ids are private, as a basic security.measure

packets

• version specification
• a byte set to 0 to end a string field
• no more connection type
• user list update request
• alias name change package
• error 250: server exploded

aka installation guide

When a profanity has been sent to the server, all the users will receive a message by (you know who) which praises the sender

this is a bug request but also it's supposed to be an example on how to use templates

version
development branch update-Min-async

Describe the bug
An UnsupportedOperationException is thrown during tests

To Reproduce
Execute PCPManager_Test.initialize()

Expected behavior
the mock server should answer with a serverexploded error packet and close the connecition

I think the api should be designed following the strategy ( or as i call it, plug-and-play ) design pattern.
This means an hypotetical PCPServerSideConnection would envelop all of the other useful components, wich could be changed at runtime based on software version or even be shared between multiple consumers.

Server-side, this could imply developing a middleware to manage computational resources.
In the #3 scheme this could be positioned between Network and Socket layer.

I will draw a scheme when i have time.

version
PCP-Min.a.2

Describe the bug
The PCPVariablePayloads.toBytes() method does not correctly format packets whose length is a multiple or greater than 2048

To Reproduce
Steps to reproduce the behavior:

1. Create a PCPVariablePayload packet its length is greater than or equal to 2048
2. Examinate the byte arrays length
3. The values will not match what is expected

Expected behavior
If an istance of a packet which extends PCPVariablePayloads have total length:

• multiple of 2048: the last packet has to containt only the header and eventually some delimitators, (packet length < 2048 obviously)
• greater than 2048: the last packet length has to be equal to header length + last part of messagge + delimitators, (packet length < 2048 obviously)

• initial implementation
• testing & debugging

I'm thinking how to convert our ArrayList<String> listOfUSers in json format.
If I would use an external library (GSON), how can i do?

Implementing PCPSocket

I propose tests for the interpreter could go in the same test classes of their respective PCPdata, as some sort of reverse engineering test.

Middlewere tests i think are gonna be the real complicated ones.

• middlewere
• interpreter