camel-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Hadrian Zbarcea <>
Subject Re: Failing Unit Tests
Date Thu, 03 Jul 2008 14:22:39 GMT

There are still some tests failing.  If I were to build a new RC  
today, I'd have to disable them.

As far as the "point of a unit test" goes, there are different schools  
of thought.  In TDD for instance writing a unit test first (which will  
fail until the feature is implemented) is not only accepted, but  
encouraged.  As long as there is an issue to track it, I am personally  
ok with disabling a test.  And actually we *do* plan to put fixes in,  
we only need to prioritize.  And we highly welcome community  
contributions in any shape or form, from wiki updates, comments on  
issues, to unit tests and even fixes.

Many thanks for your contributions,

On Jul 3, 2008, at 4:48 AM, Willem Jiang wrote:

> Got all the tests in camel-script passed.
> Please add more unit tests so we can keep camel evolving ;)
> Willem
> Aaron Mulder wrote:
>> I would prefer to see Bamboo fail at the end.  But as far as
>> commenting out tests goes, it seems to me the point of a unit test is
>> to show whether something works, and commenting-out every unit test
>> that fails really means the unit tests as a whole are pretty
>> worthless.  I would rather comment out Bamboo and let the unit tests
>> accurately reflect the state of the project.  :)
>> As far as the scripting issues and 1.4/1.5, I don't have a big
>> personal interest in whether the issues get fixed, it just seems kind
>> of cheesy that the Web site talks all about these different scripting
>> languages and how they work great for this and that, but then you try
>> them, and they turn out to be totally broken.  Perhaps if we don't
>> plan to put fixes in, we should update the Web site pages for Jython
>> and JavaScript (and I haven't tried PHP and some of the others yet!)
>> to indicate that they don't actually work.  At least we won't be
>> misleading anyone that way.  (Of course, it may just be my scripting
>> syntax that's the problem, but I was hoping someone had worked with
>> those features and would know :)
>> Thanks,
>>       Aaron
>> On Wed, Jul 2, 2008 at 9:15 PM, Willem Jiang  
>> <> wrote:
>>> Since your are meeting with a known issue which filled in JIRA, so  
>>> I comment
>>> out the unit test to let bamboo keep running.
>>> If you take a look at bamboo's building log , you will find that  
>>> it will not
>>> build the remain modules if it get a test failure.
>>> Maybe we need change the bamboo's build scrip to let it use -fae  
>>> ( fail at
>>> end) options.
>>> BTW, We are going to release Camel 1.4.0 this week, if we can't  
>>> get your
>>> issue fixed on time , I'd like to comment it out for moving it  
>>> into the
>>> 1.5.0.
>>> The JIRA for your issue could keep the unit test in our mind :)
>>> Willem
>>> Aaron Mulder wrote:
>>>> So I've been wrangling with various scripting languages in Camel
>>>> expressions.
>>>> I think I've determined that JavaScript and Python are broken,  
>>>> because
>>>> the expressions when invoked via JSR-223 always evaluate to null or
>>>> false.  It looks like that might be fixed by changing the way the
>>>> script is invoked.
>>>> Anyway, James asked me to put in unit tests to demonstrate the
>>>> problem, so I did.  Unfortunately, there are a number of unit  
>>>> tests in
>>>> the camel-script module that were previously failing and commented
>>>> out, for reasons that aren't clear to me (though I didn't really
>>>> investigate).
>>>> The problem is, someone immediately commented out my new unit tests
>>>> too, because they caused the Bamboo build to fail.  And that's a
>>>> problem because now they're just in the mess of "failing unit tests
>>>> being ignored" instead of "failing tests demonstrating a problem  
>>>> that
>>>> needs to be fixed".
>>>> So I guess I'd like to ask that the Bamboo maintainer not comment  
>>>> out
>>>> these tests, even though it will cause the build to fail if tests  
>>>> are
>>>> enabled.  If that's not acceptable, if all failing unit tests will
>>>> just be commented out, then it seems to me there's little reason to
>>>> *write* unit tests.  But I'm open to suggestions on this one.
>>>> Thanks,
>>>>     Aaron

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