[
https://issues.apache.org/jira/browse/THRIFT-3028?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jens Geyer resolved THRIFT-3028.
--------------------------------
Resolution: Fixed
Fix Version/s: 0.9.3
Assignee: Jens Geyer
Committed.
{quote}
This was not caught by tests, because the cross-tests appear not to include the
Cocoa platform.
{quote}
Guess why I created THRIFT-3100 ... ;-) Jokes aside, you're completely and
utterly right, that topic should absolutely be addressed in near future.
> Regression caused by THRIFT-2180
> --------------------------------
>
> Key: THRIFT-3028
> URL: https://issues.apache.org/jira/browse/THRIFT-3028
> Project: Thrift
> Issue Type: Bug
> Components: Cocoa - Library
> Affects Versions: 0.9.3
> Environment: Reported against ios-arm64.
> Reporter: Jeremy W. Sherman
> Assignee: Jens Geyer
> Labels: endianness
> Fix For: 0.9.3
>
> Attachments: 3028.patch
>
>
> Changes made in THRIFT-2180 to TBinaryProtocol broke little-endian message
> reading. Message-writing was also affected.
> This was not caught by tests, because the cross-tests appear not to include
> the Cocoa platform.
> A minimal patch is attached to address this.
> I wrote a local test using a StreamTransport, which lets me use NSStream to
> read/write via in-memory buffers easily (or file handles - the existing
> transports probably should have been written in terms of this standard
> Foundation abstraction) to verify that write/read of a message agree on how
> to read-write.
> [~gurubook]: Would you be so kind as to apply the attached patch, and verify
> it addresses the issue you reported on THRIFT-2180?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
