On Mon, Oct 8, 2012 at 10:14 PM, Alex Huang <[email protected]> wrote:
>
>
>> -----Original Message-----
>> From: David Nalley [mailto:[email protected]]
>> Sent: Monday, October 08, 2012 6:07 PM
>> To: [email protected]
>> Subject: Re: [ASF40][QA] AWSAPI packging remarks
>>
>> On Mon, Oct 8, 2012 at 7:43 PM, Chip Childers <[email protected]>
>> wrote:
>> > On Mon, Oct 8, 2012 at 7:38 PM, Frank Zhang <[email protected]>
>> wrote:
>> >>>
>> >>> Not being a packaging guy, I don't have a strong opinion about this 
>> >>> issue.
>> >>> However, is the consensus that we have enough of a problem here that
>> >>> it needs to be addressed prior to a release?
>> >>
>> >> Personally I think it needn't.
>> >> And I even  think awsapi should be a separate project, though this is
>> >> little off topic
>> >
>> > Thanks Frank...
>> >
>> > Any other opinions?
>> >
>> > Anyone want to take a crack at resolving the AWS API packaging issue?
>>
>>
>> Frank:
>>
>> My question is what's the next version impact of leaving things like the
>> awsapi/client %post symlinks in place? Even if we correct this in 4.next - 
>> will
>> this clean up cleanly? I will try and look at ascertain what the impact is
>> tomorrow, but am happy to have someone beat me to it. If it is something
>> that has little or no impact I am happy for us to just file bugs. If it ends 
>> up
>> being ugly, we should fix it before we try and release.
>>
> David,
>
> This is actually in current versions of CloudStack so, unlike the 
> /cloud/agent symlink, it's not something new created due to 4.0 release.  
> Because it's in previous versions of CloudStack, we have to cleanup anyways 
> if we decide not to use this "hack".  And removing it would mean we need to 
> add in code now to cleanup previous versions of links.  I think it's better 
> to file a bug and QA the proper cleanup and deploy in the next release.
>
> --Alex


Yeah - I don't think it's so awful that it hurts us, but it does point
to a problem in the AWSAPI code that we are getting around with a
packaging hack.
I filed https://issues.apache.org/jira/browse/CLOUDSTACK-293 to deal
with this in 4.1 as I am worried that the changes will cause more
uncertainty and potential problems than 'fixing it' at this stage in
the release cycle.

--David

Reply via email to