curator-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From John Vines <vi...@apache.org>
Subject Re: EnsurePath constructor
Date Mon, 16 Sep 2013 19:59:41 GMT
Yeah, EnsurePath seems to have a fairly limited use case, now that I think
about it. It does make the code a lot tidier, at the very least.


On Mon, Sep 16, 2013 at 3:43 PM, Jordan Zimmerman <
jordan@jordanzimmerman.com> wrote:

> I see - yeah that is pretty ugly. That's some of the oldest code in
> Curator as well. Since that class was created, I actually prefer
> creatingParentsIfNeeded() to EnsurePath. creatingParentsIfNeeded() only
> executes when ZK throws NoNodeException. I wonder if EnsurePath should be
> deprecated.
>
> -JZ
>
> On Sep 16, 2013, at 2:37 PM, John Vines <vines@apache.org> wrote:
>
> > Sorry, didn't mean constructor but the ensure() call. With the
> construction through newNamspaceAwareEnsurePath (which could probably stand
> to be documented better, as the EnsurePath doc page mentions the existance
> of a method and links to the CuratorFramework docs which have no mention of
> it whatsoever), it does seem strange to have to provide a
> CuratorZookeeperClient considering you're creating it through the
> CuratorFramework. I just wonder if there should be an extension/wrapper in
> curator-framework or higher that is more code friendly.
> >
> >
> > On Mon, Sep 16, 2013 at 3:00 PM, Jordan Zimmerman <
> jordan@jordanzimmerman.com> wrote:
> > 1. It's in the curator-client project, not curator-framework (i.e. it's
> low level)
> > 2. Because of namespaces, users should call
> CuratorFramework.newNamespaceAwareEnsurePath() instead of directly
> allocating an EnsurePath object.
> >
> > -JZ
> >
> > On Sep 16, 2013, at 1:55 PM, John Vines <vines@apache.org> wrote:
> >
> > > Is there a particular reason the EnsurePath constructor takes in
> > > CuratorZookeeperClient instead of a CuratorFramework like a lot of the
> > > other curator items? I thought about making a ticket, but I figured it
> was
> > > better to ask first.
> >
> >
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message