Perf are not the same for sure but the advantage is to not go through HTTP
layer:

- control of threads
- control of state "as usual"

etc...

local is the more enhanced version of the embedded behavior arquillian
promotes. Servlet reaches 80% only of it with arq 1.x - which is already
insane but sometimes tests are easier to write without these contraints.



Romain Manni-Bucau
@rmannibucau <https://twitter.com/rmannibucau> |  Blog
<http://rmannibucau.wordpress.com> | Github <https://github.com/rmannibucau> |
LinkedIn <https://www.linkedin.com/in/rmannibucau> | Tomitriber
<http://www.tomitribe.com>

2015-10-05 5:13 GMT-07:00 Alex Soto <asot...@gmail.com>:

> But what is the advantage of using local instead of http? I am sure that
> the performance is going to be pretty similar.
>
> El dl., 5 oct. 2015 a les 14:01, Romain Manni-Bucau (<
> rmannibu...@gmail.com>)
> va escriure:
>
> > You can see it the opposite as well ie you go through http where you can
> > not desire it.
> >
> > 50-50 case i think.
> > Le 5 oct. 2015 13:52, "John D. Ament" <john.d.am...@gmail.com> a écrit :
> >
> > > The fundamental problem (if I had to guess) is that let's say I'm
> > testing a
> > > REST API (heck even using the REST extension), persistence extension
> > > doesn't know that my request went through since I'm controlling it, so
> > its
> > > call backs won't take effect.  By running embedded, you're more likely
> to
> > > not be inside the same transaction, which is one of the things being
> > banked
> > > on I'm sure.
> > >
> > > John
> > >
> > > On Mon, Oct 5, 2015 at 3:26 AM Alex Soto <asot...@gmail.com> wrote:
> > >
> > > > If there are bugs/improvements let s do it. I am thinking in terms of
> > > doing
> > > > the life of developers easier and in fact changing the protocol to
> > > servlet
> > > > dos not introduce (a priori) any back incompatibility but a lot of
> > > > extensions are going to work with embedded mode.
> > > >
> > > > Alex.
> > > >
> > > > El dg, 4 oct 2015 a les 7:53 Romain Manni-Bucau <
> rmannibu...@gmail.com
> > >
> > > va
> > > > escriure:
> > > >
> > > > > Hi Alex
> > > > >
> > > > > Local protocol allows nicer start so very tempted to keep it like
> > that.
> > > > > Servlet protocol has few bugs/issues and is not what you target by
> > > > default
> > > > > with embedded adapters IMO.
> > > > > Le 4 oct. 2015 01:25, "John D. Ament" <johndam...@apache.org> a
> > écrit
> > > :
> > > > >
> > > > > > Hey Alex,
> > > > > >
> > > > > > To be honest, I find the protocol section of arquillian the most
> > > > > confusing
> > > > > > for new users to pick up.  Why should they care what the protocol
> > > > > > transferring the test data is using?  I remember at the
> beginning,
> > > AS7
> > > > > had
> > > > > > a JMX protocol.  This made tons of sense, since it didn't impact
> > the
> > > > > > running application.  No one understood why their tests stopped
> > work
> > > > > > though!  Truth be told, the fact that most people end up relying
> on
> > > > > > servlet, without even realizing that its the arquillian runtime
> > > > starting
> > > > > an
> > > > > > HTTP request makes it confusing to understand scope.  Basically,
> > > users
> > > > > end
> > > > > > up relying on the servlet protocol to start their HTTP request
> for
> > > > them,
> > > > > > instead of starting an HTTP request for their test invocation.
> > > > > >
> > > > > > Just curious, is this out on discourse somewhere? I can continue
> my
> > > > whine
> > > > > > fest there :-)
> > > > > >
> > > > > > John
> > > > > >
> > > > > > On Sat, Oct 3, 2015 at 4:58 PM Alex Soto <asot...@gmail.com>
> > wrote:
> > > > > >
> > > > > > > Hello mates,
> > > > > > >
> > > > > > > I send this email to start a discussion for a change on TomEE
> > > > > Arquillian
> > > > > > > Adapter.
> > > > > > >
> > > > > > > The other day, because of the book (Arquillian In Action), I
> > > checked
> > > > > the
> > > > > > > protocol used by different adapters. I noticed that only Weld
> > > adapter
> > > > > and
> > > > > > > TomEE embedded adapter are using the local protocol,all other
> > ones
> > > > are
> > > > > > > using the Servlet 3.0 protocol. And maybe you are asking why
> this
> > > is
> > > > > > > important? Well the problem is that a lot of extensions
> requires
> > > this
> > > > > > > protocol to work, for example Arquillian Persistence extension
> or
> > > > > > Arquillan
> > > > > > > Warp extension. In case a user wants to use  embedded TomEE and
> > one
> > > > > > > extension that requires servlet protocol, it would get a
> cryptic
> > > > > > exception.
> > > > > > > Of course there is a workaround and it is overriding the
> default
> > > > > protocol
> > > > > > > by using protocol special tag in arquillian.xml and adding the
> > > > > > > servlet-protocol dependency.
> > > > > > >
> > > > > > > I think that it would be easier for everyone if all adapters
> > works
> > > in
> > > > > the
> > > > > > > same way. With the release of TomEE 7 I think it could be a
> good
> > > > > > > opportunity to make this change in the Arquillian embedded
> > adapter
> > > as
> > > > > > well
> > > > > > > and align with the rest of the containers.
> > > > > > >
> > > > > > > The change to be done is very simple in terms of code (Adding
> one
> > > > > > > dependency and change one String of the adapter).
> > > > > > >
> > > > > > > What do you think? From the point of view of Arquillian
> community
> > > it
> > > > > > would
> > > > > > > be the logic step.
> > > > > > >
> > > > > > > Alex.
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>

Reply via email to