On Fri, Feb 7, 2025 at 3:06 AM G. Branden Robinson <g.branden.robin...@gmail.com> wrote: > At 2025-02-06T23:23:57-0600, Dave Kemper wrote: > > The info file documents it the same way it documents everything else: > > any attribute associated with an environment has this association > > explicitly stated; attributes without this explicit statement are > > global. > > The practice mirrors CSTR #54, where a capital "E" in the synopsis of a > request marks it as being (in my term) "environmental". It's not quite > obvious that anything that isn't environmental is global in AT&T troff, > because there's no annotation for a global property, and the same field > of the synopsis is used to list properties of diversions, default > scaling units, break implication, typesetting-specific operations, and a > few other things. > > (I'll opine here that I dislike the documentary practice of using a > simple list (compounded by the use of implicit nulls) to enumerate > orthogonal parameters. I gather that one can project R^n onto the real > line, but the mechanism used to achieve any such mapping is seldom > obvious to the reader.)
Despite its flaws, the CSTR #54 approach does impart the information better to readers than the groff manual's, I think. When there's a column that might or might not contain an E--even if the column is overloaded with other information as well--the absence of the E makes it clearer to readers that the listed property is global than does the absence of a specific sentence in the info manual's nontabulated prose. The info manual's approach also requires essentially the same sentence to be repeated numerous times throughout. And because info manuals are designed to be perused nonsequentially, even the addition of an introductory statement that properties are global unless stated to be environmental only partly ameliorates the first issue. > Is the selection of hyphenation language a hyphenation parameter? > > Setting aside your knowledge of formatter behavior leading to this > thread, what would your assumption have been based on those words? I'm not sure I can answer that without bias. I *think* I would have considered settings like .hy and .hys to clearly be hyphenation parameters, and things set by .hw and .hla to be less clear: they strike me more as the framework to which the parameters apply (although even that isn't as straightforward as I had at one time expected: my original complaint in http://savannah.gnu.org/bugs/?53196 was partly that the .hy setting does not apply to breakpoints defined by .hw--a fact that was undocumented until Werner added a paragraph to manual in response to that bug report).
