> is it safe to divert into a string variable? In a way, roff doesn't really distinguish between "strings", "macros", and "diversions". They all share the same namespace and can be accessed both via ".xx" at the beginning of an input line and as "\*[xx]". The results, however, will depend on what the "variable" currently contains.
> That means instead of: > > .br > .di A > ...text... > .br > .di > .A > > use something like: > > .br > .di A > ...text... > .br > .di > ...\*A... *Diverting* text means that you store *formatted* material in the variable. During formatting, for example, spaces become fixed. Formatted material is not intended to be reprocessed again, e.g., in fill mode. Although it is possible, if you do so you're likely to get uneven spacing. For single words it might be okay (except if the diversion was formatted with a nonzero indent). If you intend to reprocess saved text, you should normally save that text unprocessed as a macro: .br .de A ...text... .. .A more text or a string: .br .ds A ...text... .A \" remember: A has no newline, so this is stupid more text The difference between ".de" and ".ds" is that ".ds" does not store a terminating newline in A, which is why stuff defined with ".ds" should really be accessed as \*A. This also helps to preserve the structure of the input with regard to lines. E.g., if you have an "emboldening" macro, regular text .B bold text regular text and a *macro* A containing multiple lines, then regular text .B \*A regular text will give you only the first line of A in bold, and the rest in regular (and an empty line afterwards).
