I didn't see the implementation itself, but sometimes i experience something similar in unit tests running inside Eclipse and in command-prompt environment. When Strings are put in a HashSet (*not* TreeSet), the string order (during the iteration) is different (even though the VM is the same).
Don't know if it's relevant to this scenario...
----- Original Message ----
From: Lowell Kirsh <lowellk_at_gmail.com>
To: users_at_jaxb.dev.java.net
Sent: Friday, October 6, 2006 2:12:06 PM
Subject: Re: JAXB not repeatable?
The hard thing is that this happens so rarely that I've been uable to reproduce it. But I will definitely try to send yo something if I can figure out how to replicate it.
On 10/5/06,
Kohsuke Kawaguchi <Kohsuke.Kawaguchi_at_sun.com> wrote:
Lowell Kirsh wrote:
> By the way, I forgot to mention that this is using jaxb2.
>
> On 10/5/06, *Lowell Kirsh* <lowellk_at_gmail.com
> <mailto:
lowellk_at_gmail.com>> wrote:
>
> Not that this helps, but I have had similar problems. I have a type
> defined in my schema:
>
> <xs:complexType name="numericList">
> <xs:choice maxOccurs="unbounded">
> <xs:element ref="number" />
> <xs:element ref="numberRange"/>
> </xs:choice>
> </xs:complexType>
>
> The java code that it produces is sometimes
> getNumbersAndNumberRanges() and other times it generates
> getNumberRangesAndNumbers(). This is without changing the schema at
> all.
>
> If anyone knows how to fix this, please let me know.
Really? That sounds like a serious bug, as the latter is actual
violating the spec. We generally pay a lot of attention to those things.
A test case would be really great.
--
Kohsuke Kawaguchi
Sun Microsystems kohsuke.kawaguchi_at_sun.com