Re: svn commit: r1001368 - in /ofbiz/trunk/applications/content/data: DemoBlogEntryData.xml DemoBlogMiscData.xml

[Scott Gray-2]

On 1/10/2010, at 8:40 AM, Adam Heath wrote:

> On 09/30/2010 02:35 PM, Scott Gray wrote:
>> On 1/10/2010, at 8:30 AM, Adam Heath wrote:
>>
>>> On 09/30/2010 01:03 PM, Scott Gray wrote:
>>>> On 1/10/2010, at 6:04 AM, Jacques Le Roux wrote:
>>>>
>>>>> From: "Adam Heath"<[hidden email]>
>>>>>> On 09/29/2010 09:11 PM, Scott Gray wrote:
>>>>>>> Thanks Hans and no worries, I don't think it was as big of a deal as it was made it out to be.
>>>>>>>
>>>>>>> If people think it might be a good idea, we could always alter the way the data loader works so that loading errors result in a failed build.  This would cause buildbot to notify us as soon as the problem occurs and allow the committer to fix the issue while the work is still fresh in their mind.
>>>>>>
>>>>>> Isn't that always the case?  ant run-install fails, so the tester on buildbot fails.
>>>>>
>>>>> Right, that how I knew the issue was there. It's sent on dev ML BTW.
>>>>>
>>>>
>>>> No that's incorrect, data load problems never result in a failed build at the moment.
>>>
>>> Yes, I see that now.  The fix has to be involved with the container/component subsystems.  Working on it...
>>
>> It's pretty straight forward, just throw a ContainerException and then check for the return status in the ant target.
>
> Except it'd be nice that if a container fails to load, all previously loaded/started containers are shutdown before just returning.

That would be nice, currently the ContainerLoader does nothing with the boolean return value, we could change that.

[Jacques Le Roux]

Scott Gray wrote:

> On 1/10/2010, at 6:04 AM, Jacques Le Roux wrote:
>
>> From: "Adam Heath" <[hidden email]>
>>> On 09/29/2010 09:11 PM, Scott Gray wrote:
>>>> Thanks Hans and no worries, I don't think it was as big of a deal as it was made it out to be.
>>>>
>>>> If people think it might be a good idea, we could always alter the way the data loader works so that loading errors result in
>>>> a failed build.  This would cause buildbot to notify us as soon as the problem occurs and allow the committer to fix the issue
>>>> while the work is still fresh in their mind.  
>>>
>>> Isn't that always the case?  ant run-install fails, so the tester on buildbot fails.
>>
>> Right, that how I knew the issue was there. It's sent on dev ML BTW.
>>
>
> No that's incorrect, data load problems never result in a failed build at the moment.

Indeed, I mixed Hans' issues with Anil's

Jacques
 
> Regards
> Scott