Why is the logic of the MultiAgentFormatter different between the openai and other model?

[ChPi]

Java version uses appendRoleAndName while Python version only appends the name: accumulated_text.append(f"{msg.name}: {block['text']}")

[ChPi]

agentscope-java/agentscope-core/src/main/java/io/agentscope/core/formatter/openai/OpenAIConversationMerger.java

Line 148 in f153fb0

appendRoleAndName(textBuffer, roleLabel, agentName);

openai实现和其他不一样

agentscope-java/agentscope-core/src/main/java/io/agentscope/core/formatter/dashscope/DashScopeConversationMerger.java

Line 95 in f153fb0

textAccumulator.append(name).append(": ").append(tb.getText()).append("\n");