On Thu, Feb 06, 2025 at 04:05:36PM +0000, Werner LEMBERG wrote: > > > https://www.gnu.org/software/texinfo/manual/texinfo/html_node/HTML-Xref-8_002dbit-Character-Expansion.html > > Thanks, but I don't think this is a valid explanation. 'o' is > definitely *not* the NFC of 'ö'! AFAICS, the above warning message > points to a bug in `texi2any`.
Not the part about the regular rules, but about transliteration for file names: .... To handle such cases, texi2any offers the --transliterate-file-names command line option. This option enables transliteration of node names into ASCII characters for the purposes of file name creation and referencing. The transliteration is based on phonetic principles, which makes the generated file names more easily understandable. This is the default. > > What should happen is that the two nodes are merged in one HTML file > > and the redirection file should work for both, as long as the > > processor generating the manual uses the same conventions. > > Looking into the created file `bogen_html/Bogen.html` I only see code > for handling 'Bogen'. There is no redirection code for handling > 'Bögen'. Indeed, I was too hasty and your example is not with nodes but anchors. In that case, there is only a redirection file created and it will only point to one anchor (the first one, if I recall well). I do not think that it is a bug given that --no-transliterate-file-names can be used to generate a redirection file with B_00f6gen.html name. If --no-transliterate-file-names is used both for this manual and a manual that would refer to Bögen, the result will be what you expect. There is a trade-off between more pleasing file names, with transliteration, but with clashing anchors redirection files, and no clashing file names but less pleasing file names. Maybe a solution would be to add the possibility to state in xhtmlxref.cnf that a manual is generated with --no-transliterate-file-names such that other manuals referring to it use the same convention as obtained with --no-transliterate-file-names. -- Pat
