Hi Eliot, We made no modifications to the DITA OT processing (added no patches, changed no code, stylesheet or build file). So like you guessed, the files are brought down using the Oxygen supported protocols (HTTP, FTP, ZIP or custom plugin-installed protocols). They are not being brought down as a pre-processing (all at once) but the mechanism leads to a similar result (for a more technical description I can give you more details if you are interested in a direct email). The user agent we are sending for HTTP is the usual one:

Oxygen XML Editor/CURRENT_VERSION

but I see no difficulty in modifying its value slightly on our side if the DITA OT processing is the one which initiated the HTTP download. We'll contact you when this gets implemented. Regards, Radu Radu Coravu <oXygen/> XML Editor, Schema Editor and XSLT Editor/Debugger http://www.oxygenxml.com On 10/28/2011 5:33 PM, Eliot Kimber wrote:

I haven't had a chance to test this new feature yet, but I'm wondering is it simply a pre-process that accesses all the dependencies and makes a local copy with pointers rewritten as necessary or is it actually modifying the Ant processes to allow direct operation on HTTP-based resources?

If the former, what user agent string does Oxygen use when making the HTTP requests?

I'm thinking about this in the context of our content management systems (RSuite CMS, DocZone, RSuite Cloud), where currently we have built-in facilities to do map export and then run the Toolkit but where all the resources are URL-accessible and so could be processed directly.

However, in most cases, some amount of transformation is required to be applied to the data as it comes out of the repository, either to filter out things like CMS-specific metadata or to rewrite pointers to reflect the location as exported.

But we also have a facility for imposing transformations into object access requests, configurable by conditions including the user agent. So, for example, if Oxygen is requesting resources specifically in order to then run the Toolkit on them, given a distinguishing user agent string I could configure transforms to condition the data as it comes out without having implement a completely separate export action. That would be very handy.

On the other hand, if it is the Toolkit, not Oxygen that is requesting the resources, that might require different transformation logic, not sure.

So really I'm just curious about what's really going on under the covers.

Cheers,

E.