Follow-up Comment #3, bug #61450 (project groff): [comment #0 original submission:] > Conversation causing this bug report occurred on the groff > mailing list in 2021-11 under the subject ‘Two trivial questions’.
For easy clicking, the relevant thread starts at http://lists.gnu.org/r/groff/2021-11/msg00043.html Another aspect has occurred to me since that thread: I wonder if the present behavior is by design, much like how unrecognized macros are silently (by default) ignored. The advantage of such leniency is that a document designed for a more-featured roff can be processed by a less-featured one, with perhaps some loss of fidelity but otherwise intact. For example, a document prepared for groff might use the .tkf request to adjust letterspacing, but could still be processed with a historical roff that lacks this request; the desired letterspacing would of course not be reflected, but the document would largely render fine otherwise. Similarly, groff's .ss request takes two parameters, but this is an extension to CSTR#54 troff, where it took only one. But thanks to the lack of parameter-count checking, a two-parameter .ss will be handled without complaint by a CSTR#54-only troff, with only slight changes in sentence spacing to show for it. So if, down the road, some future groff 1.28 expands the .po request to take two additional optional parameters, documents that use this expanded functionality can still run without error on groff 1.23, just missing whatever bells and whistles 1.28 added. This isn't an argument against such warnings being activable with the proper switch (as undefined-macro warnings currently are), but it's possibly an argument against their being on by default, and probably an argument against one day upgrading the condition to an error. _______________________________________________________ Reply to this item at: <https://savannah.gnu.org/bugs/?61450> _______________________________________________ Message sent via Savannah https://savannah.gnu.org/
