You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@thrift.apache.org by "Dmytro Shteflyuk (Jira)" <ji...@apache.org> on 2026/08/12 11:00:00 UTC

[jira] [Updated] (THRIFT-6142) Ruby HeaderTransport does not limit unframed messages

     [ https://issues.apache.org/jira/browse/THRIFT-6142?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Dmytro Shteflyuk updated THRIFT-6142:
-------------------------------------
    Fix Version/s: 0.25.0

> Ruby HeaderTransport does not limit unframed messages
> -----------------------------------------------------
>
>                 Key: THRIFT-6142
>                 URL: https://issues.apache.org/jira/browse/THRIFT-6142
>             Project: Thrift
>          Issue Type: Bug
>          Components: Ruby - Library
>            Reporter: Dmytro Shteflyuk
>            Assignee: Dmytro Shteflyuk
>            Priority: Major
>             Fix For: 0.25.0
>
>          Time Spent: 50m
>  Remaining Estimate: 0h
>
> h3. Problem
> Ruby HeaderTransport supports unframed Binary and Compact clients by passing reads directly to the underlying transport after detecting the protocol. The configured max_frame_size is enforced for Header and framed clients, but it is not applied to these supported unframed messages.
> h3. Client impact
> Applications configuring max_frame_size can still receive an unframed Binary or Compact message larger than that limit. The setting therefore behaves differently depending on the automatically detected client transport type.
> h3. Reproduction
> Testing on master commit {{a9663bc6661a5dd1d99d629e1f269c1907592a1a}} produces:
> {code:ruby}
> require "thrift"
> {
>   "binary" => Thrift::BinaryProtocol,
>   "compact" => Thrift::CompactProtocol
> }.each do |label, protocol_class|
>   buffer = Thrift::MemoryBufferTransport.new
>   writer = protocol_class.new(buffer)
>   writer.write_message_begin("legacy", Thrift::MessageTypes::CALL, 1)
>   writer.write_struct_begin("Args")
>   writer.write_field_stop
>   writer.write_struct_end
>   writer.write_message_end
>   payload = buffer.read(buffer.available)
>   transport = Thrift::HeaderTransport.new(
>     Thrift::MemoryBufferTransport.new(payload)
>   )
>   limit = payload.bytesize - 1
>   transport.set_max_frame_size(limit)
>   reader = protocol_class.new(transport)
>   name = reader.read_message_begin.first
>   reader.skip(Thrift::Types::STRUCT)
>   reader.read_message_end
>   puts "#{label}: decoded=#{name.inspect} " \
>        "bytes=#{payload.bytesize} max_frame_size=#{limit}"
> end
> {code}
> {noformat}
> binary: decoded="legacy" bytes=19 max_frame_size=18
> compact: decoded="legacy" bytes=11 max_frame_size=10
> {noformat}
> h3. Expected behavior
> HeaderTransport should apply max_frame_size to each unframed Binary or Compact protocol message. Messages at or below the configured limit should remain supported, messages above it should raise TransportException::SIZE_LIMIT, and the budget should reset between sequential messages.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)