Return-Path: X-Original-To: apmail-hive-dev-archive@www.apache.org Delivered-To: apmail-hive-dev-archive@www.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 7307D17A23 for ; Wed, 14 Jan 2015 19:39:38 +0000 (UTC) Received: (qmail 683 invoked by uid 500); 14 Jan 2015 19:39:39 -0000 Delivered-To: apmail-hive-dev-archive@hive.apache.org Received: (qmail 607 invoked by uid 500); 14 Jan 2015 19:39:39 -0000 Mailing-List: contact dev-help@hive.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: dev@hive.apache.org Delivered-To: mailing list dev@hive.apache.org Received: (qmail 508 invoked by uid 500); 14 Jan 2015 19:39:39 -0000 Delivered-To: apmail-hadoop-hive-dev@hadoop.apache.org Received: (qmail 475 invoked by uid 99); 14 Jan 2015 19:39:39 -0000 Received: from arcas.apache.org (HELO arcas.apache.org) (140.211.11.28) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 14 Jan 2015 19:39:39 +0000 Date: Wed, 14 Jan 2015 19:39:39 +0000 (UTC) From: "Ashutosh Chauhan (JIRA)" To: hive-dev@hadoop.apache.org Message-ID: In-Reply-To: References: Subject: [jira] [Commented] (HIVE-9376) Add back some APIs which changed from 0.13 for Spark SQL MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-JIRA-FingerPrint: 30527f35849b9dde25b450d4833f0394 [ https://issues.apache.org/jira/browse/HIVE-9376?page=3Dcom.atlassian.= jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D14277= 535#comment-14277535 ]=20 Ashutosh Chauhan commented on HIVE-9376: ---------------------------------------- I have a concern with such a change. Such kind of patch adds tech debt on p= roject. We are forced to maintain multiple signatures of same method becaus= e some other project used an internal api. Such patching on our side will e= ncourage those projects to make more usage of internal apis vs using docume= nted public stable apis. I think contract is if any project wants to use an= internal api, they should be ok with its breakage in next hive release. At= that point, they can update the signature to latest hive version, if they = chose to support it. Else, they can use well established shimming technique= if they want to continue using internal apis with multiple hive version su= pport. (Hive itself has done this for long time for hadoop). Providing guar= antees for backward compatibility of any method an external project might c= hose to use will dissuade Hive devs to make any changes in hive codebase. P= ure refactoring patches which improves code will be looked down upon, since= who knows who might use method getting refactored.=20 > Add back some APIs which changed from 0.13 for Spark SQL > -------------------------------------------------------- > > Key: HIVE-9376 > URL: https://issues.apache.org/jira/browse/HIVE-9376 > Project: Hive > Issue Type: Sub-task > Reporter: Brock Noland > Assignee: Brock Noland > Fix For: 0.15.0 > > Attachments: HIVE-9376.1.patch > > > As [discussed on HIVE-9364|https://issues.apache.org/jira/browse/HIVE-936= 4?focusedCommentId=3D14276100&page=3Dcom.atlassian.jira.plugin.system.issue= tabpanels:comment-tabpanel#comment-14276100] some APIs were changed which b= roke SparkSQL. This change adds them back, marks them as Deprecated, Unstab= le, and in-use by SparkSQL. -- This message was sent by Atlassian JIRA (v6.3.4#6332)