On Mon, Oct 12, 2026 at 09:30:18AM +0200, Patrice Dumas wrote: > On Sun, Oct 11, 2026 at 10:03:36PM +0200, Patrice Dumas wrote: > > Therefore another customization variable could be needed. Another > > possibility would be to use a string instead a boolean for > > AUTO_SECTION_NODE, with 3 possibilities > > * AUTO_SECTION_NODE=add: add nodes in Info and HTML only if not > > ambiguous (otherwise add nodes with postpended [+1+] and [+2+]), > > and for other formats add targets only if not ambiguous. The default. > > * AUTO_SECTION_NODE=warn: always add a target and warn if there are > > duplicate > > * AUTO_SECTION_NODE=silent: always add a target if there are duplicate and > > do > > not warn > > This is not good, it is not possible with this scheme to differenciate > between HTML/Info (added nodes) and other formats (added targets only). One > possibility would be, instead, to add another boolean variable: > NO_WARN_DUPLICATE_ADDED_SECTION_TARGET > > It could probably also be possible to have a string option with 4 > possibilities, but I can't come up with good names.
If I understand correctly, you want a way to configure whether warnings are given for duplicate automatically added section targets in both the cases when they are made into nodes (AUTO_SECTION_NODE=1) and when they are not (AUTO_SECTION_NODE=0). It makes sense to add another boolean variable as you suggest. We could probably come up with a shorter name: maybe WARN_DUPLICATE_SECTION_TARGET? I would suggest that if we had two variables affecting this feature, AUTO_SECTION_NODE and WARN_DUPLICATE_SECTION_TARGET, they should operate in an independent way. The defaults could be different for different output formats (I don't know why that would be, but that is what you wanted to do for Info and HTML), but setting one shouldn't change the value of the other. I think this means the default for Info and HTML is AUTO_SECTION_NODE=1 and WARN_DUPLICATE_SECTION_TARGET=0. For other formats it is AUTO_SECTION_NODE=0 and WARN_DUPLICATE_SECTION_TARGET=1. I'd be happy with WARN_DUPLICATE_SECTION_TARGET=0 for all output formats, although, as I believe we discussed, this should affect only a few manuals, so outputting a warning is an acceptable choice. When we document these variables and the "shadow target" functionality, I would like to avoid being specific about whether references to duplicate targets work or not. So we shouldn't explicitly allow for this with different keyword values for one of the customisation variables, as you initially proposed above. The functioning of WARN_DUPLICATE_SECTION_TARGET may be slightly different depending on the setting of AUTO_SECTION_NODE but I believe the purpose of this variable would be clear enough to users in either case.
