gmane.comp.java.jmock.user

Re: jmock 2.6 compatibility with Java 7

Steve Freeman — 2013-04-20T09:41:50

It doesn't build with 7, but the compiled jars work with 7.

S

On 19 Apr 2013, at 18:10, Prasanna wrote:

Steve Freeman

Winner of the Agile Alliance Gordon Pask award 2006
Book: http://www.growing-object-oriented-software.com

+44 797 179 4105
Twitter: < at >sf105
Higher Order Logic Limited
Registered office. 2 Church Street, Burnham, Bucks, SL1 7HZ. 
Company registered in England & Wales. Number 7522677




---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

jmock 2.6 compatibility with Java 7

Prasanna — 2013-04-19T17:10:42

Hi,

Is jmock 2.6 compatible with java 7 ?

Thanks,
Prasanna


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Fwd: Unexpected behaviour with package private classes - IllegalAccessError

Duncan Jones — 2013-02-21T13:26:17

(Sent to developer list by mistake - please see below)

---------- Forwarded message ----------
From: Duncan Jones <duncan-VgTxFcESRta8IPqWsmCfhg< at >public.gmane.org>
Date: 21 February 2013 13:06
Subject: Unexpected behaviour with package private classes - IllegalAccessError
To: dev-sXN/XchZ9OexIXFVlbCvtR2eb7JE58TQ< at >public.gmane.org

Hi,

I've encountered some strange behaviour when using package private
classes within expectations. The proxy class gets an
IllegalAccessError when referring to a package private class instance
in a return value.

I've attached a minimal working Maven project ("example-project.zip"),
which demonstrates the behaviour. I'm unclear whether this is expected
or not.

Some notes:

 - The issue does not occur if the method in question returns null
(see testSuccess() in my example test file).

 - If I convert ExampleInterface to be package private, the error does
not occur.

Kind regards,

Duncan

---------------------------------------------------------------------
To unsubscribe from th

Re: doesn't compile with Java 7. Breaking change?

Steve Freeman — 2012-12-31T13:18:12

Thanks for offering, but I don't think that'll do much good. The same rules apply for Java 8 and beyond. We've been pushing the boundaries of what can be done with generics and it looks like we found an edge case. In the end, the underlying problem is the brokenness of the language, in this case: type erasure and primitive types.

S.

On 30 Dec 2012, at 22:42, James Abley wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: doesn't compile with Java 7. Breaking change?

James Abley — 2012-12-30T22:42:59

That sounds nasty. Java being how it is, have Oracle really released a
breaking change like that? I'm tempted to raise on one of the JDK lists.

On Dec 30, 2012 10:23 PM, "Steve Freeman" <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:
type system so that it doesn't know what to do with all the overloads of
the with() method in ArgumentConstraintPhrases. If we delete the overloads
for the primitive types to make the code compile, then it fails at runtime
because we cannot return null as a primitive type and we don't have a
runtime way to find out what type we're supposed to be returning.
Expectations.WithClause where we specify the return type in the method name
for primitives, for example:
(2.7). This new API should still compile and work with Java 6, even if it's
more clunky.

doesn't compile with Java 7. Breaking change?

Steve Freeman — 2012-12-30T22:23:10

It turns out that JMock doesn't compile with Java7. They've changed the type system so that it doesn't know what to do with all the overloads of the with() method in ArgumentConstraintPhrases. If we delete the overloads for the primitive types to make the code compile, then it fails at runtime because we cannot return null as a primitive type and we don't have a runtime way to find out what type we're supposed to be returning. 

The best we can think of is to do something based on Expectations.WithClause where we specify the return type in the method name for primitives, for example:

  oneOf(foo).blah(with.intIs(lessThan(2)));

Unfortunately, this is a breaking change.

Unless someone has a better idea, we'll do this in the next release (2.7). This new API should still compile and work with Java 6, even if it's more clunky.

Thoughts?

S.



---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

jmock.org, now including javadoc

Steve Freeman — 2012-12-19T18:20:09

so, I had to hide away in the mountains, but I believe I've finally managed to get the javadoc working for released builds on http://www.jmock.org

S.
---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Duncan Jones — 2012-12-01T22:18:55

Toby - yes, that's what I was suggesting. I'm rather fond of Maven,
however, so I suspect my views are a little biased...


On 27 November 2012 09:52, Toby <toby.weston-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

Re: Due date for 2.6.0 (revisited)

Steve Freeman — 2012-11-28T08:46:29

thanks



---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Toby — 2012-11-27T19:30:22

Hey,

Started to capture actions as issues on GitHub (
https://github.com/jmock-developers/jmock-library/issues), that way we can
be clear about what we're doing and can assign people to help out.

They're prefixed "[GitHub Migration] : Release process" but feel free to
delete/edit/abuse what's there so far.

More to follow I'm sure.

On 27 November 2012 12:02, Steve Freeman <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:

Re: Due date for 2.6.0 (revisited)

Steve Freeman — 2012-11-27T12:02:44

I'd prefer B. I'd like to keep the core of jmock clean of the big dependencies...

S.
---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Toby — 2012-11-27T10:18:05

As it goes, I suspect it's less effort to continue with ant and deploy using maven. To be clearer, I think of it like this

A) jmock-library project : uses Ant : build and release (which means tag) and deploy to jmock.org/dist

B) jmock-packaging project : uses maven : takes artefacts from above and deploys to maven central

Moving A) to use maven may or may not be a big deal. Maven usually ends up being a big deal in my experience. Effort to try is ...?

Toby


On 27 Nov 2012, at 09:10, Steve Freeman <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Toby — 2012-11-27T09:52:54

I think Duncan's talking about the option to build in the jmock-library project using maven (removing Ant) and removing the packaging project (and release and deploy directly from Maven). Is that right Duncan?


On 27 Nov 2012, at 09:10, Steve Freeman <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Steve Freeman — 2012-11-27T09:10:10

Not sure I understand that. My understanding is that, for maven, we generate some artifacts, keep the pom up to date, and then upload. Is there more than that?

S

On 27 Nov 2012, at 08:11, Duncan Jones wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Steve Freeman — 2012-11-27T08:05:17

it's the new one. 4.11

S.

On 27 Nov 2012, at 00:34, elias-aaUjpfuAgBPQT0dZR+AlfA< at >public.gmane.org wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Duncan Jones — 2012-11-27T08:11:56

Reading the other mails in this thread, it seems like quite some effort is
required to support a Maven release whilst using ant as a build system. Are
you categorically ruling out a move to building JMock via Maven or are you
open to cajoling? :-)

Duncan


On 23 November 2012 10:02, Steve Freeman <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:

Re: Due date for 2.6.0 (revisited)

2012-11-27T00:34:12

Is JMock 2.6.0 the one with Hamcrest 1.3/describeMismatch support?

Thanks!

On 2012-11-25 02:44, Andreas Pieber wrote:

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Andreas Pieber — 2012-11-25T08:44:12

I would say so; at least my +1 :-)

Kind regards,
Andreas

On Sat, Nov 24, 2012 at 12:12 PM, Steve Freeman <steve-/6t1Ty3A97Vaa/9Udqfwiw< at >public.gmane.org> wrote:

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

Steve Freeman — 2012-11-24T11:12:50

so, is sonatype.org the answer?

S.

On 23 Nov 2012, at 12:55, Andreas Pieber wrote:


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

Re: Due date for 2.6.0 (revisited)

James Shaw — 2012-11-23T18:09:17

There's a maven plugin to automate the last few steps:
http://www.sonatype.com/books/nexus-book/reference/staging-sect-managing-plugin.html

I have yet to use them myself, though.
On 23 Nov 2012 12:55, "Andreas Pieber" <anpieber-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

Re: Due date for 2.6.0 (revisited)

Andreas Pieber — 2012-11-23T12:55:30

Not nexus-like; it's a nexus they have running :-) . The manual step
isn't great but it isn't a real problem neither. I'm deploying all my
OS software via oss.sonatype.org and it's IMHO not a show stopper. At
least the only real option to get away from codehaus.

Kind regards,
Andreas

On Fri, Nov 23, 2012 at 1:44 PM, Toby <toby.weston-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email