>We propose an experiment based on RFC 3933 allowing, in addition to >ASCII text as a normative input/output format, PDF as an additional >normative output format. There are a lot of different formats called PDF. There are PDF 1.1, 1.2, 1.3, and 1.4. There's the new PDF/A archival profile along with a variety of other industry-specific PDF/x profiles . And there are a whole lot of files produced by alleged PDF generators that don't actually conform to any version of the PDF spec. (Often they depend on non-standard fonts that happened to be installed on the author's computer.) Among valid PDFs, do you include PDFs that are coded to prohibit text extraction? How about PDFs that are just bitmap scans of printed documents, like the PDF versions of some early RFCs from the 1970s? As we all know, one of the reasons that ASCII text has stood the test of time is that its definition is stable and well-understood, so it is at no risk of becoming unreadable due to losing the programs needed to decode it. I think that PDF/A may be well enough defined to be an adequate archival format, but just "PDF" is way too vague. R's, John _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf