Well, you can't compile the entire project with JDK 1.4 right now.
Using multiple JVMs to compile seems like it would be considerably
more complicated than using JDK 1.5 for the whole build process with
the backwards-compatible JDK 1.4 jar specified.
On Aug 21, 2007, at 10:58 AM, Phill Moran wrote:
Can't you trigger on the JDK version setup in the POM. SO if you
are compiling
on 1.4 one profile is active and if 1.5 another and so on. Is there
a specific
need to trigger on a jar file
-----Original Message-----
From: Marc Prud'hommeaux [mailto:[EMAIL PROTECTED] On
Behalf Of Marc
Prud'hommeaux
Sent: August 21, 2007 10:54 AM
To: [email protected]
Subject: Re: svn commit: r567962 -
/openjpa/trunk/openjpa-jdbc/src/main/java/org/apache/openjpa/jdbc/
sql/DB2Diction
ary.java
Also, you can have the validation automatically happen if your ~/.m2/
settings.xml looks like:
<settings>
<profiles>
<profile>
<id>java14-validate</id>
<activation><activeByDefault>true</activeByDefault></
activation>
<properties>
<java14.jar>/System/Library/Frameworks/
JavaVM.framework/Versions/1.4.2/Classes/classes.jar</java14.jar>
</properties>
</profile>
</profiles>
</settings>
The exact location of "java14.jar" is platform and installation
specific, which
is why we can't automatically set it (unless anyone has any clever
suggestions
for figuring out the path of different JVM's installation in a
robust manner).
On Aug 21, 2007, at 10:41 AM, Patrick Linskey wrote:
On 8/21/07, Kevin Sutter <[EMAIL PROTECTED]> wrote:
Nice catches, Patrick. These 1.5-specific items keep slipping
through.
I've been talking with Mike and Marc about this -- maybe there's
something more automated that can be done to help prevent these type
of changes from slipping through.
So Marc and I both catch these things because we're running automated
build systems that run with the elaborate arguments required to link
against the 1.4 jars. One possibility would be for you guys to
compile
with those same args, but that's a bit of a pain. Another would b e
for me to finally get around to making my CI system available to
others who want access to it. I'll look into that in the next day or
so.
The other change you caught with the unbalanced parens kind of
scares
me.
We ran both the regression bucket and JPA TCK against DB2 without
this getting caught. Maybe this particular path isn't executed by
those tests.
I am checking with Teresa to see if we need some new testcases for
this.
As Teresa mentioned, the code itself was fine; the balancing was just
being done across a method boundary.
-Patrick
--
Patrick Linskey
202 669 5907
