Mailing-List: contact jspwiki-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: jspwiki-dev@incubator.apache.org
Received-SPF: pass (nike.apache.org: domain of andrew.r.jaquith@gmail.com
 designates 209.85.221.191 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type:content-transfer-encoding;
        b=DHpEfn46zVcobwbHvhbxkl+Vaexx/SywKRiVdxUoB/zvoxPpPZUF9GRjcLwdw7j6IW
         RQu6Gv8SG1LIS/e0HDHs1TNPBv1vKzerd0jd6c/I+mtP2Q9mc/PJIiXeAFjSCDgzVt8h
         mImQ6wZ9b/HVmCienOeLdfXZ8DueAN6D9+yMk=
MIME-Version: 1.0
In-Reply-To: <9EC83A3C-FE3F-41DB-9185-E075F3BEF28B@ecyrd.com>
References: <b90102160912010544j29205f67ha3cf2e660fe42e4f@mail.gmail.com>
	 <9EC83A3C-FE3F-41DB-9185-E075F3BEF28B@ecyrd.com>
Date: Tue, 1 Dec 2009 12:59:27 -0500
Message-ID: <b90102160912010959sabbc1dsff321e076b778b2@mail.gmail.com>
Subject: Re: Failing DeleteActionBean tests
From: Andrew Jaquith <andrew.r.jaquith@gmail.com>
To: jspwiki-dev@incubator.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

The easy way to test this hypothesis, which is an interesting one,
would be to unhook the ReferenceManager event listener and see if it
makes a difference. The RefMgr tests will fail, but perhaps some of
the other failing tests will succeed.

Generally, though, we should definitely do one of two things: either
create a separate "updater" JCR session for the ReferenceManager event
listener, and have it do its own saves, OR remove all save()s from
ReferenceManager entirely. Right now it does saves in the context of
the caller's JCR session, which doesn't feel right.

Andrew

On Tue, Dec 1, 2009 at 11:25 AM, Janne Jalkanen
<Janne.Jalkanen@ecyrd.com> wrote:
>
> The test seems correct. =A0There seems to be something fairly odd going o=
n
> with respect to versioning; for example, ContentManagerTest.deleteAllPage=
s()
> fails because it deletes all pages ok, but at the last instant it recreat=
es
> the pages as empty.
>
> My guess is that it was a bug in ContentManager that priha 0.6.x revealed=
;
> and I also have a nagging suspicion it has something to do with
> ReferenceManager, as it seems to me that the pages are recreated by the
> event handlers as a side effect. =A0This could happen if the page is firs=
t
> deleted, then save()d, then a event handler adds a property to it and
> recreates the Node as a sanity check or something. Haven't yet had time t=
o
> peek into it very deeply.
>
> /Janne
>
> On Dec 1, 2009, at 15:44 , Andrew Jaquith wrote:
>
>> Janne, can I get a quick sanity check on something?
>>
>> =A0 =A0 =A0// Save two versions of the test page
>> =A0 =A0 =A0 m_engine.saveText( "Test", "This is the first version" );
>> =A0 =A0 =A0 m_engine.saveText( "Test", "This is the second version" );
>>
>> =A0 =A0 =A0 // Make sure they both saved ok
>> =A0 =A0 =A0 WikiPage v1 =3D m_engine.getPage( "Test", 1 );
>> =A0 =A0 =A0 WikiPage v2 =3D m_engine.getPage( "Test", 2 );
>> =A0 =A0 =A0 assertNotNull( "Did not save page Test, v1!", v1 );
>> =A0 =A0 =A0 assertNotNull( "Did not save page Test, v2!", v2 );
>> =A0 =A0 =A0 assertEquals( "This is the first version",
>> m_engine.getPureText( v1 ).trim() );
>>
>> The last assertion is failing. When WikiEngine.getPureText(v1) is
>> called, it is returning "This is the second version", which seems
>> wrong . It should return the first version, right?
>>
>> Back when the trunk tests all ran clean (aka "the good old days"),
>> this test worked also. So it seems to me that some recent changes
>> caused the results to change. So was the test wrong all along, or is
>> there a bug somewhere?
>>
>> Andrew
>
>