[Building Sakai] Proposal: upgrade dependency versions in 1.0 kernel

[Steve Swinsburg]

> and close on 6months of local production use

Normally I don't advocate new additions to a stable branch, but if these upgrades alleviate problems and have been in production for 6 months then +1. 

cheers,
Steve



On 11/05/2010, at 1:37 AM, David Horwitz wrote:

> See bellow
> 
> On 05/10/2010 05:07 PM, Matthew Buckett wrote:
>> On 10 May 2010 13:43, David Horwitz <david.horwitz at uct.ac.za> wrote:
>> 
>>> For the 2.7 release a number of kernel dependencies where upgraded some
>>> time ago. Testing of 2.7 has indicated that these all seem safe. The
>>> list can be seen here:
>>> 
>>> http://jira.sakaiproject.org/browse/KNL-105
>>> 
>>> an includes:
>>> hibernate
>>> spring
>>> javax.mail
>>> 
>> I think this broke some stuff in UserDirectoryService (MimeEncode.base64()).
>> 
>> 
> 
> There's a patch for uds in the issue linked to the javax upgrade - that
> would be added at the same time.
>>> cglib
>>> commons-lang
>>> commons-fileupload
>>> commons-logging
>>> 
>>> I would suggest that these now be applied to the 1.0.x kernel series for
>>> 2.6, with the exception of the spring upgrade. Any objections?
>>> 
>> To me this is a risk without any clear benefits. I would be happy for
>> the versions in the 2.6.x to remain the same so that there aren't any
>> surprises for people running the stable software.
>> 
>> If there is a bug in one of the libs which affects Sakai that can only
>> simply be fixed by upgrading then it would seems the sensible route to
>> upgrade that lib to the newer version.
>> 
>> 
> 
> There have definitely been people affected by bugs in some of these
> libraries and a number have performance issues). In the past we have
> been badly stung by not updating these. Notably in a file upload bug
> that caused the jvm to consume all system resources. Im proposing this
> because I believe now the risk is small enough (based on the 2.7 testing
> and close on 6months of local production use) that this is outwayed by
> the possible gains in subtle bugs, egde cases and performance issues.
> 
> D
> 
> _______________________________________________
> sakai-dev mailing list
> sakai-dev at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
> 
> TO UNSUBSCRIBE: send email to sakai-dev-unsubscribe at collab.sakaiproject.org with a subject of "unsubscribe"