impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Armstrong (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-4008: don't bake in hash table and hash join pointers
Date Tue, 13 Sep 2016 19:49:25 GMT
Tim Armstrong has posted comments on this change.

Change subject: IMPALA-4008: don't bake in hash table and hash join pointers

Patch Set 3:

File be/src/exec/hash-table.h:

PS3, Line 300: ExprValuesHash
> CurExprValuesHash()  [see below for motiviation]

PS3, Line 303: SetExprValuesHash
> Let's rename this to more closely match cur_expr_values() rather than match

PS3, Line 309: cur_expr_values_null
> comment explaining that there is one byte per null value (i.e. these are us

I'm not actually sure why we need this for codegen, since bool and uint8_t should both be
represented as int8 internally in LLVM, but I don't want to get distracted digging into that,
so I left a TODO.

There's some weirdness with vector<bool> where it uses bitfields but otherwise C++ bools
should be 8 bits - maybe that was the reason.

PS3, Line 403: in
> the first time I read this, I read it as this functions stores the values t
Makes sense - done.

PS3, Line 403: in
> same

PS3, Line 413: in
> same

PS3, Line 431: This will be replaced by
             :   /// codegen.
> nit: move this sentence to end of the comment so it's standardized with sim

PS3, Line 438: with 'cur_expr_values_'
> "to compare against the current row."  (since it also uses cur_expr_values_

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: Ie353666dbb5c958f0094d169306fe930ec3014c5
Gerrit-PatchSet: 3
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Tim Armstrong <>
Gerrit-Reviewer: Dan Hecht <>
Gerrit-Reviewer: Michael Ho
Gerrit-Reviewer: Michael Ho <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message