Re: Groff's -mm Indexing

[G. Branden Robinson]

Hi Damian,

At 2023-07-01T01:20:44+1000, Damian McGuckin wrote:
> .INITI P
> .IND item0
> .IND item1
> .IND item2
> .INDO

The foregoing gives good exercise to new groff mm indexing system
validation logic, which will show up early in the next release cycle.

$ ./build/test-groff -rRef=1 -mm -z EXPERIMENTS/damian-index.mm
m.tmac:EXPERIMENTS/damian-index.mm:1: error: INITI: invalid index type 'P'

Fix that by switching the type to 'N' (or another valid choice
documented in the groff_mm(7) man page)...

$ ./build/test-groff -rRef=1 -mm -z EXPERIMENTS/damian-index.mm
m.tmac:EXPERIMENTS/damian-index.mm:2: error: IND: cannot record index entry 
when not on a page

I put that diagnostic in today.

Fix that...

$ ./build/test-groff -ww -rRef=1 -mm -z EXPERIMENTS/damian-index.mm
.\" IND item0   1
.\" IND item1   1
.\" IND item2   1
troff:EXPERIMENTS/damian-index.mm:6: warning: macro 'INDO' not defined

(You'll notice a "-ww" crept into my command line.)

The macro is spelled "INDP".  Fix that...[1]

$ ./build/test-groff -ww -rRef=1 -mm -z EXPERIMENTS/damian-index.mm
.\" IND item0   1
.\" IND item1   1
.\" IND item2   1
.\" Index: damian.ind

Splendid!

I've also added a unit test to validate this.  It promptly exposed a bug
I'd introduced last month (on my private branch only).

Regards,
Branden

[1] This required yet another change to stop the package from shutting
    all warnings off when the `Ref` register is non-zero.  In my opinion
    that is unhelpful and unnecessary.  Diagnostics on stderr are
    already distinguishable by dint of format; the macro package is
    careful (and has been for many years) to write out this non-error
    information prefixed with a *roff comment character sequence, which
    is a pretty unlikely match for a diagnostic message produced by
    anything even vaguely conforming to the GNU Coding Standards.

damian-index.mm
Description: Text Data

signature.asc
Description: PGP signature