[bug #66174] [mdoc] `Bl` macro's `-width` option poorly documented

[G. Branden Robinson]

Follow-up Comment #6, bug #66174 (group groff):

At 2024-10-15T04:20:15-0400, Bjarni Ingi Gislason wrote:
> Follow-up Comment #5, bug #66174 (group groff):
> 
>   The report is valid as it shows a regression.

I don't agree.  The input document uses incorrect syntax for what it's
trying to do.

The same "regression" would occur if the formatter implementation
widened its integer type to 64 bits; "1234567890123" would then neither
wrap nor saturate.  ((2^63)-1 is 9223372036854775807.)

You will search in vain for a specification of *roff that mandates a
specific bit width for its integer type or specifies any of the 3 means
of handling overflow: trapping, wrapping, or saturating.

https://dl.acm.org/doi/10.1145/363235.363259

> Installed "nroff" (version 1.23.0) and "mandoc" render the man page
> fine.

The document got lucky with those formatters.  It should be changed to
say what it means: either prefix the integer literal with `\&` to make
the argument a string, or express the width as a numeric expression
(measurement), `13n`.



    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?66174>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

signature.asc
Description: PGP signature