I run a fairly large service for reviewing conference submissions, almost all in PDF, with several ten thousand submissions each year. You'd be amazed how much broken PDF is out there, produced by all kinds of tools. Older versions of Microsoft Office and various "free" PDF conversion tools routinely fail to include fonts, for example, thus, violating PDF/A requirements. Others break PDF so that text can't be cut-and-pasted. We don't even try for PDF/A compliance, just a much narrower set of requirements imposed by the major academic EE/CS publishers (IEEE and ACM). Unfortunately, I have not found a reasonable online tool (service) that produces correct PDF 100% of the time. Adobe Acrobat works most of the time for the desktop, but is pretty expensive. The Mac version of Word seems to have a high success ratio. http://www.freepdfconvert.com/ works frequently. Etc. We still do a fair amount of hand-tweaking and tool-trying, so if you truly want 100% PDF/A compliance, be prepared not just for buying a fairly expensive tool chain to check this (we use Callas PDFtools for compliance checking), but also technical support staff to deal with the 1% or so of files that won't convert well. NSF also does online conversion from Word to PDF. It kind of, mostly works. (Such tools tend to be available for Windows servers only, unfortunately, and are certainly not free. These tools typically script an Office client to do the conversion, since that's the only sure way to get something that approximates the PDF output that Office produces. Some use OpenOffice, but this seems to sometimes cause rendering problems.) Thus, requiring authors to upload "random" PDF isn't unreasonable, in my view, but requiring PDF/A or doing server-side PDF conversion raises the bar significantly, for all parties in the process. Henning On Nov 17, 2011, at 12:52 PM, David Morris wrote: > > > On Thu, 17 Nov 2011, Robinson Tryon wrote: > >> >> If authors take on the responsibility of creating and verifying the >> fidelity of exported versions, then I think everything will be peachy. >> What can we do to encourage this practice? > > Start by compensating them for the work required to conform to this > requirement? > > The IETF is a bunch of volunteers, most of whom carve out time for > participation from time needed to satisfy other responsiblities. > > The authors already have issues with the basic tool set and compatiblity > with the computing environments needed for other life responsiblities. > Now you want to dictate a new step. > > Adding a new tool/process is absurd. If you have a solution that actually > works for everyone without adding much to their time burden, test it, > demonstrate it with your own materials, etc. > _______________________________________________ > Ietf mailing list > Ietf@xxxxxxxx > https://www.ietf.org/mailman/listinfo/ietf > _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf