Hmmm...
I just cribbed HashMap from the Sun sources, and changed the equals()
and hashCode() calls. I'll try to get a chance to look at it today,
but no promises.
Anyone want to run a diff against 1.4's IdentityHashMap?
(or just copy it from the JDK into Jaxen?)
-bob
On Fri, 3 May 2002, Eddie Mc Greal wrote:
> point taken - didn't think of that
> However there is still a problem - I don' think the IdentityHashMap works
> correctly.
>
> The following doesn't produce the correct result
> count(PLAY/ACT/SCENE[1] ) returns 1
> when I take out the identity map I get 5 but then all the ancestor test
> fail! (obvious to me now )
> So, I'll commit my version of the DefaultLocationPath as I'm pretty sure it
> is correct - can you look at the IdentityMap issue - all the ancestor test
> are currently broken
> Eddie
>
> ----- Original Message -----
> From: bob mcwhirter <[EMAIL PROTECTED]>
> To: Eddie Mc Greal <[EMAIL PROTECTED]>
> Cc: <[EMAIL PROTECTED]>
> Sent: Friday, May 03, 2002 3:51 PM
> Subject: Re: [Jaxen] DefaultLcationPath
>
>
> > On Fri, 3 May 2002, Eddie Mc Greal wrote:
> >
> > > after furter investigation I found the following
> > > the reason for the change (1.8 to 1.9) was the following xpath
> > >
> > > count(/PLAY/ACT/SCENE[1]) should be 5 and not 1.
> > > However, the fix for this problem (made by xcut) is incorrect.
> > > The problem is the use of the unique map to filter out identical nodes
> but /PLAY/ACT/SCENE produces several SCENE nodes. Only adding the first one
> leads to the bug that count only finds 1 instead of 5.
> > > I've removed the unique map (any objections? why was it used in the
> first place?) and checked it in. All test pass including
> count(/PLAY/ACT/SCENE[1])
> >
> > The IdentityHashMap?
> >
> > That's there, because of things involving '..'.
> >
> > /PLAY/ACT/..
> >
> > That selects the ACTs, obviously, and then their parents.
> >
> > Typically, a play has 5 ACTs, so then we're looking at 5 instances of
> > the PLAY parent. Being a node-set (instead of a node-bag), we have to
> > eliminate the duplicate instances of PLAY.
> >
> > The Map, if its the one I'm thinking of, though, should be an
> > IdentityHashMap, which would allow 5 different ACT elements in it, as
> > long as they were not the same instance. (It uses the identity hashCode
> > and ==, instead of hashCode() and equals()).
> >
> > So, I guess the test we need is that:
> >
> > count(/PLAY/ACT/SCENE/../..) == 1
> >
> > or
> >
> > count(/PLAY/ACT/SCENE/..) == 5
> >
> > -bob
> >
> >
> > _______________________________________________________________
> >
> > Have big pipes? SourceForge.net is looking for download mirrors. We supply
> > the hardware. You get the recognition. Email Us: [EMAIL PROTECTED]
> > _______________________________________________
> > Jaxen-interest mailing list
> > [EMAIL PROTECTED]
> > https://lists.sourceforge.net/lists/listinfo/jaxen-interest
>
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply
the hardware. You get the recognition. Email Us: [EMAIL PROTECTED]
_______________________________________________
Jaxen-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jaxen-interest
