- From: Alessandro Vernet <avernet@orbeon.com>
- Date: Wed, 29 Mar 2006 18:24:46 -0800
- To: public-xml-processing-model-wg@w3.org
On 3/23/06, Innovimax SARL <innovimax@gmail.com> wrote: > The idea is to allow backtracking of the pipeline (manage the pipes by > taking the outputs of it) and looking if condition changes so the > processor could regenerate the content > > So S2 which is the last for example has a test which could tell to the > processor (which could just be, "has the file timestamp of A changed > ?" Mohamed, Assuming we have a pipeline "A -> S1 -> B -> S2 -> C", in which S1 and S2 are two steps, document A is transformed in document B by S1, and document B transformed in document C by S2, then you are saying that you would the pipeline language not to prevent an implementation from being able to detect when running the pipeline that the pipeline has already been executed before, that A has not changed since then, that S1 and S2 have no side effect, and that consequently there is no need to actually to run the steps S1 and S2, as they would generate the same output documents. Is this way to put it consistent with the idea you had in mind? If it is, I for one am favorable to include this in our list of use cases. Alex -- Blog (XML, Web apps, Open Source): http://www.orbeon.com/blog/
Received on Thursday, 30 March 2006 02:24:51 UTC