Return-Path: Delivered-To: apmail-incubator-cayenne-user-archive@locus.apache.org Received: (qmail 46151 invoked from network); 29 Sep 2006 15:03:52 -0000 Received: from hermes.apache.org (HELO mail.apache.org) (209.237.227.199) by minotaur.apache.org with SMTP; 29 Sep 2006 15:03:52 -0000 Received: (qmail 76558 invoked by uid 500); 29 Sep 2006 15:03:51 -0000 Delivered-To: apmail-incubator-cayenne-user-archive@incubator.apache.org Received: (qmail 76541 invoked by uid 500); 29 Sep 2006 15:03:51 -0000 Mailing-List: contact cayenne-user-help@incubator.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: cayenne-user@incubator.apache.org Delivered-To: mailing list cayenne-user@incubator.apache.org Received: (qmail 76532 invoked by uid 99); 29 Sep 2006 15:03:51 -0000 Received: from idunn.apache.osuosl.org (HELO idunn.apache.osuosl.org) (140.211.166.84) by apache.org (qpsmtpd/0.29) with ESMTP; Fri, 29 Sep 2006 08:03:51 -0700 X-ASF-Spam-Status: No, hits=0.0 required=5.0 tests= Received: from [193.138.94.41] ([193.138.94.41:36802] helo=mx4.unvienna.org) by idunn.apache.osuosl.org (ecelerity 2.1.1.8 r(12930)) with ESMTP id E9/F5-13110-2D53D154 for ; Fri, 29 Sep 2006 08:03:48 -0700 Received: from unvsmtp2.unvienna.org (unvsmtp2.unvienna.org [172.17.252.136]) by mx4.unvienna.org (8.12.11/8.12.11) with ESMTP id k8TF3cJk011329 for ; Fri, 29 Sep 2006 17:03:39 +0200 In-Reply-To: <0317438A-1328-44D1-94B7-8D402178D001@objectstyle.org> Subject: Re: Problems with MySql id type with AUTO_INCREMENT To: cayenne-user@incubator.apache.org X-Mailer: Lotus Notes Release 7.0.1 January 17, 2006 Message-ID: From: Tobias SCHOESSLER Date: Fri, 29 Sep 2006 17:03:35 +0200 X-MIMETrack: Serialize by Router on unvsmtp2/VIENNA/UNO(Release 7.0.1|January 17, 2006) at 29/09/2006 17:03:38 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII X-Spam-Rating: minotaur.apache.org 1.6.2 0/1000/N I am a bit confused now ... so java.lang.Long PKs are known to not work at all with cayenne 1.2 no matter which database is used? I had the impression everything worked fine when using MySQL directly with BIGINT, Long and database side AUTO_INCREMENT. My last bug I ran into was only when I connected through the sequoia jdbc driver / mysql jdbc driver chain when the test case broke again ... Anyway I am flexible with the pk column types but because of this MySql jdbc driver bug mentioned before http://lists.mysql.com/java/6383 - returning meta data of type java.lang.Long for any AUTO_INCREMENT column defined - I thought I was forced to use some database type that consistently maps as java.lang.Long for the PK to not have to deal with changing pk types on creation and retrieval . That's why i tried INT UNSIGNED and BIGINT. Is your advise still correct for this ? Andrus Adamchik To cayenne-user@incubator.apache.org Friday, 29 cc September 2006 15:57 Subject Re: Problems with MySql id type with AUTO_INCREMENT Please respond to cayenne-user@incu bator.apache.org Tobias, If you have flexibility redefining your columns, I suggest mapping a PK as INTEGER until we implement Long PK support for real (http:// issues.apache.org/cayenne/browse/CAY-329). As far as I can tell in the your latest case you are hitting the int pk limitation of PK generator. This issue is pretty high on our priority list by the way, so Long pks should become an option soon. Andrus On Sep 29, 2006, at 9:51 AM, Tobias SCHOESSLER wrote: > Andrus, thanks a lot you for looking into this. > > That is quite a pile of bugs I am running into here. First this > MySQL JDBC > driver bug returning java.lang.Long for any auto-increment column, > then > cayenne treating 'INT UNSIGNED' as an Integer. Now issues with > org.apache.cayenne.access.jdbc.ColumnDescriptor and SQLTemplate. > > To get this running for me I understand that I only have to pick > the right > datatype for my PKs so all the bugs are bypassed. Looking at > http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-type- > conversions.html > I thought BIGINT should be unambigiously treated as a > java.lang.Long for > all bugs mentioned. > > Actually it worked, I tried setting the PKs to mysql type BIGINT > and the > test case mentioned below passes. So for the record this is a > solution for > using AUTO_INCREMENT PK generation with cayenne 1.2 and MySQL. > > But... > > If you recall my ultimate goal was to use a setup in which MySql is > used in > a Sequoia/CJDBC cluster - the reason I did not use AUTO_PK_SUPPORT > as LOCK > was not supported by Sequoia ... > > Connecting through the cjdbc/sequoia driver to the database above, > again > breaks the test case. Sequoia returns java.lang.Integer for the BIGINT > column ... : -[ > > I guess this is rather something to address in a sequoia forum or > is there > a chance you add a sequoia setup in your test cases? > > thanks > > Tobias > > > > > Andrus Adamchik > > le.org> To > cayenne- > user@incubator.apache.org > Thursday, > 28 cc > September 2006 > 17:44 > Subject > Re: Problems with MySql id type > with AUTO_INCREMENT > Please respond to > cayenne-user@incu > bator.apache.org > > > > > > > > Just got some time to test it, and the problem seems to be in > org.apache.cayenne.access.jdbc.ColumnDescriptor constructor... It is > getting confused by MySQL ResultSetMetaData returning JDBC type for > such column as INTEGER. > > I created a new Jira issue with description of this problem: > > http://issues.apache.org/cayenne/browse/CAY-672 > > I'll commit a fix to 3.0, as 1.2 users can have some code relying on > the old behavior and we want to keep 1.2 stable. > > Andrus > > > On Sep 26, 2006, at 7:11 AM, Tobias SCHOESSLER wrote: >> Andrus, >> >> I am referring to http://issues.apache.org/cayenne/browse/CAY-639 >> and our >> discussion below. >> >> I am using cayenne 1.2.1 now. I mapped all AUTO_INCREMENTED primary >> keys of >> sql type unsigned int as BIG_INT in cayenne. >> >> Cayenne correctly returns DataObjects with a primary key of type >> Long when >> using SelectQuery, when creating new objects or using getToXXX >> navigation >> to retrieve the object . >> >> But when I use SQLTemplate to retrieve the same DataObject cayenne >> still >> returns a pk of type Integer ... >> >> These are my test cases: >> >> >> public void testDocumentPkViaSelectQuery() { >> List docList = >> dataContext.performQuery(new >> SelectQuery(TldDocument.class, null)); >> assertEquals(Long.class, >> docList.get(0).getObjectId().getIdSnapshot().get >> (TldDocument.DOCUMENT_UID_PK_COLUMN).getClass()); >> >> } >> >> public void testDocumentPkViaSQLTemplate() { >> >> String sqlSelect = " select * "; >> String sqlFrom = " from TLD_DOCUMENT doc "; >> >> SQLTemplate query = new SQLTemplate >> (TldDocument.class, >> sqlSelect + " " + sqlFrom); >> List docList = >> dataContext.performQuery(query); >> >> assertEquals(Long.class, >> docList.get(0).getObjectId().getIdSnapshot().get >> (TldDocument.DOCUMENT_UID_PK_COLUMN).getClass()); >> >> } >> >> testDocumentPkViaSelectQuery passes. >> >> testDocumentPkViaSQLTemplate fails. >> >> thanks, regards >> >> Tobias. >> >> >> >> >> >> Andrus Adamchik >> > >> le.org> To >> cayenne- >> user@incubator.apache.org >> Tuesday, >> 29 cc >> August 2006 15:35 >> >> Subject >> Re: Problems with MySql id >> type >> Please respond to with AUTO_INCREMENT >> cayenne-user@incu >> bator.apache.org >> >> >> >> >> >> >> >> >> I'd appreciate that. >> >> Andrus >> >> >> On Aug 29, 2006, at 5:31 PM, Tobias SCHOESSLER wrote: >> >>> ok, I got it. You want me to add a Jira entry for this? I never did >>> this >>> before ... : | >>> >>> >>> >>> >>> >>> >>> Andrus Adamchik >>> >> >>> le.org> To >>> cayenne- >>> user@incubator.apache.org >>> Tuesday, >>> 29 cc >>> August 2006 15:19 >>> >>> Subject >>> Re: Problems with MySql id >>> type >>> Please respond to with AUTO_INCREMENT >>> cayenne-user@incu >>> bator.apache.org >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> On Aug 29, 2006, at 4:16 PM, Tobias SCHOESSLER wrote: >>> >>>> But still you are right there is a bug in mysql too. If the column >>>> type is >>>> INT and not INT UNSIGNED it should return java.lang.Integer but it >>>> returns >>>> java.lang.Long instead. This was my first observation. >>> >>> Still probably worth fixing the mapping to follow the docs. >>> >>> >>>> but why does cayenne need BIGINT mapping set for the column to >>>> retrieve it >>>> as a java.lang.Long? I understand that this is unrelated to the bug >>>> in the >>>> mysql jdbc driver mentioned above. >>> >>> This is simply a workaround for the bug. BIGINT has a default JDBC >>> mapping to java.lang.Long. So this consistently produced a Long >>> during select and when reading auto-incremented value via JDBC. >>> >>> Andrus >>> >>> >>> >> >> >> >> > > > >