Tim Armstrong has posted comments on this change. ( http://gerrit.cloudera.org:8080/8438 )
Change subject: IMPALA-6137: fix text scanner split delim mem mgmt ...................................................................... Patch Set 2: (2 comments) The newest patch is my attempt at a minimal fix that addresses the edge case we discussed and works around the problem that FillByteBuffer() is overridable. http://gerrit.cloudera.org:8080/#/c/8438/2/be/src/exec/hdfs-text-scanner.h File be/src/exec/hdfs-text-scanner.h: http://gerrit.cloudera.org:8080/#/c/8438/2/be/src/exec/hdfs-text-scanner.h@143 PS2, Line 143: virtual Status FillByteBuffer(MemPool* pool, bool* eosr, int num_bytes = 0) > *sigh* I found out that this is virtual because the text scanner plugin mec I don't think I want to change the extension mechanism as part of this bug fix - filed IMPALA-6146. http://gerrit.cloudera.org:8080/#/c/8438/2/be/src/exec/hdfs-text-scanner.cc File be/src/exec/hdfs-text-scanner.cc: http://gerrit.cloudera.org:8080/#/c/8438/2/be/src/exec/hdfs-text-scanner.cc@717 PS2, Line 717: byte_buffer_last_byte_ > I couldn't see an obvious code path that would take us there. Maybe some sp I couldn't come up with a test that would cause this, but I added logic to avoid reading byte_buffer_last_byte_ when it wasn't initialised to a valid value. -- To view, visit http://gerrit.cloudera.org:8080/8438 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Iddbb5cf6acc8f0b0e0b4c205c334f21e03d06f1c Gerrit-Change-Number: 8438 Gerrit-PatchSet: 2 Gerrit-Owner: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Dan Hecht <dhe...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: anujphadke <apha...@cloudera.com> Gerrit-Comment-Date: Thu, 02 Nov 2017 20:12:47 +0000 Gerrit-HasComments: Yes