Looks pretty good to me too. My quibbles:
- Should export new_transport_factory/1 for parity with other transports.
- This may be a micro-optimization, but I think a binary representation for
the buffer might be more efficient: per
http://erlang.org/doc/efficiency_guide/binaryhandling.html#4, appending to the
back of a binary is optimized in R12B to avoid a copy. The benefit of the IO
list on the wire is clear, but since this is an in-memory buffer, we don't get
the savings of avoiding io_list_to_binary, but do pay the cost of additional
cons cells for the intermediate io-list.
- Eugene
On 1/29/09 8:26 PM, Todd Lipcon (JIRA) wrote:
[
https://issues.apache.org/jira/browse/THRIFT-126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12668741#action_12668741
]
Todd Lipcon commented on THRIFT-126:
------------------------------------
lgtm except for a nitpick - it says thrift_framed_transport in the header at
the top. No big deal but worth fixing before committing.
TMemoryBuffer for Erlang
------------------------
Key: THRIFT-126
URL: https://issues.apache.org/jira/browse/THRIFT-126
Project: Thrift
Issue Type: New Feature
Components: Library (Erlang)
Reporter: David Reiss
Assignee: David Reiss
Priority: Minor
Attachments: 0001-THRIFT-126.-erl-TMemoryBuffer-for-Erlang.patch
TMemoryBuffer is really helpful for testing protocol changes. I just
implemented it for Erlang. Patch in a moment.
