- From: Vasil Rangelov <boen.robot@gmail.com>
- Date: Mon, 1 Oct 2007 15:36:43 +0300
- To: <public-xml-processing-model-comments@w3.org>
>Now I'm confused. The input that arrives on stdin is fed to the >program as it's stdin. The outputs that it produces on stdout and >stderr are (with appropriate escaping and wrapping) returned as the >stdout and stderr outputs from the step. > >How the implementation arranges for this to happen is the >implementation's problem. You mean the input can be passed when the program is first run and it's the program's responsibility to process this input in its own fashion? Hm. I think I just learned better how the CLI really works. Thanks ^_^. >| How exactly is the input document used? Passed at once as a serialized >| document with the default serialization options? > >I had in mind that it would be passed as a serialized text file. I >guess we'd need to provide serialization options to the step. Sounds like much of a complication if you ask me, but yeah. If an XML is to be passed, such options would be needed. >| If so, what about programs >| that don't accept XML document as their input? > >Then stdin had best be bound to p:empty :-) That wasn't my point. I meant what about programs that don't accept XML document as their input, but do accept input (i.e. accept plain text input or just an input in another non XML format)? An empty binding would be like providing no input, but that doesn't deal with programs accepting text input (and especially doesn't deal with programs that would require input). The only solution I could think is for me (i.e. the pipeline author) to create another program which will translate the XML file into it's plain text equivalent (perhaps by using DOM and the wholeText property) and feed that into the program, from then on getting the result and returning it to the pipeline which would then escape it and all, but that's just needlessly tedious and inefficient. If stdin was an option or if there was some way to get the whole text contents of the XML document instead, this would sort of solve this. How about a similar XPointer (or maybe just plain XPath) syntax like the one from this other comment of mine? http://lists.w3.org/Archives/Public/public-xml-processing-model-comments/200 7Sep/0053.html If THAT was to be accepted, this issue (and many others that could occur) would be solved too. Regards, Vasil Rangelov
Received on Monday, 1 October 2007 12:37:32 UTC