[ https://issues.apache.org/jira/browse/THRIFT-2690?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon South updated THRIFT-2690: -------------------------------- Attachment: thrift-2690-c_glib-fix-seg-fault-in-testthrifttestclient.patch I've attached a second patch that fixes a seg fault in one of the c_glib unit tests, unmasked by the stricter handling of typedef'd types. > c_glib: Incorrect behaviour when serializing a map with typedef'd keys > ---------------------------------------------------------------------- > > Key: THRIFT-2690 > URL: https://issues.apache.org/jira/browse/THRIFT-2690 > Project: Thrift > Issue Type: Bug > Components: C glib - Compiler > Affects Versions: 0.9.1 > Environment: Fedora 20 64-bit on x86_64 > Reporter: Simon South > Priority: Minor > Labels: c_glib, compiler > Attachments: > thrift-2690-c_glib-fix-seg-fault-in-testthrifttestclient.patch, > thrift-2690-c_glib-resolve-typedef-when-serializing-map.patch > > > This is the same issue reported in > [THRIFT-2683|https://issues.apache.org/jira/browse/THRIFT-2683] but for > _serializing_ maps. > The compiler must resolve a typedef'd type to its underlying type before > attempting to reason about it. By failing to do this, the current c_glib > compiler will in most cases generate incorrect code for serializing a map > when either its keys or its values are of a typedef'd type. The client > receives garbage data instead of the map's actual contents. > The attached patch fixes this. It should be applied after the patch for > [THRIFT-1266|https://issues.apache.org/jira/browse/THRIFT-1266]. > No test case, yet again, but the server half of the integration test suite > will trigger this bug without the patch applied. -- This message was sent by Atlassian JIRA (v6.3.4#6332)