Those interested in the PDF output for Texinfo might like to try the new support for "PDF page labels" implemented in commit 0133695d7. With some PDF readers numbers that are actually printed on the pages as the page numbers rather than page number 1 being the very first page in the document (which would usually be the title or bastard title page). (I tested it with okular and xreader - xpdf seems to lack this feature.) I tried to test with different combinations, e.g. the contents at the end of the document. I did not support bizarre document structures such as the contents being in the middle of the document. There are at most three numbering schemes used: T-1, T-2, ... for the title pages, Roman numerals (i, ii, iii...) for the @contents and/or @shortcontents, and the regular Arabic numerals (1, 2, 3, ...) for the main content. I'd be interested if the implementation caused bad results for a plausible document structure.
