[Building Sakai] Entity References Broken During Site Import: Solution Vote (KNL-737 or SAK-16568)

[Bryan Holladay]

So it sounds like a big difference between KNL-737 and SAK16568 is KNL only
updates Site ID references and is contained completely in Kernel as a
service, while SAK-16568 updates any OLD->NEW context (Site Id, Assignment
ID, Forum ID, ect) but requires the tools to supply the refernce(old->new)
mapping and as well as the logic to convert them inside itself by
implementing the stubs.

-Bryan

On Thu, Jul 28, 2011 at 10:43 AM, Matthew Buckett <
matthew.buckett at oucs.ox.ac.uk> wrote:

> On Thu, Jul 28, 2011 at 2:07 PM, Bryan Holladay <holladay at longsight.com>
> wrote:
> > Matt can describe his and maybe we can make an educated decision here?
>
> Background:
>
> Most tools at the moment support the transfer of content from one site
> to another. When they do this however they don't update any of the
> content, so if you have a link to a PDF held in the current site's
> resource in the syllabus, when the syllabus is transferred to the new
> site the link remains pointing to the old site, if people don't have
> access to the old site they aren't able to access the PDF.
>
> Solution:
>
> So what both Bryan and I have done it have the tools re-write their
> content when they are transferring it from one site to another. What I
> did was provide a service that allows tools to have each piece of
> their content scanned for links and those links re-written and the
> content handed back to the tool. I also build list of resources that
> the content linked to and all the resources needed by that resource
> which can be automatically copied. This is so that if you are just
> importing the syllabus from another site you get a copy of the
> syllabus (with the links updated) in your new site but also any
> references to content in resources will be copied across even if you
> don't want the whole of the resources tool copied.
>
> The new service supports looking at HTML and plain text for references
> to content in resources. I have patches for all the core tools to
> support this, they are currently against 2.6.x but we are moving to
> 2.8.x over the summary so I will need to update them to close to
> trunk.
>
> I do full regexp replaces in content only if the context IDs (site
> IDs) are auto-generated so that if you allow custom site IDs we don't
> accidentally re-write content we shouldn't.
>
> I like the service solution as it means we don't end up with this
> logic duplicated across lots of tools.
>
> Issues:
>
> I don't scan the content for links to things other than resources. So
> for example, if you have a HTML file in resources which links to a
> forums discussion topic, and you are only copying the resources of a
> site, I won't automatically add the Forum to the destination site and
> create a copy of the referenced discussion topic. With our current URL
> management I don't think this would be easy todo.
>
> Side note: I'm off on holiday (for 10 days yey) in about 2 hours and
> will be away from email, so I'm not ignoring responses.
>
>   Matthew Buckett
>   VLE Developer, LTG, Oxford University Computing Services
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20110728/72e625c2/attachment.html