[Pranav-error]

Fixes #7734.

When r.stats produces no output, make_coin() parsed zero records and nothing noticed:

• the read loop ends without entering the "Unexpected output from r.stats" branch, which only catches malformed lines
• the category lists are then allocated with G_calloc(0, ...)
• collapse() starts its counter at 1, so an empty list still reports one category
• that category's value is read from the zero length allocation, which is undefined behaviour and in practice reads as 0

The result was a normal looking coincidence report for a single category 0, with EXIT_SUCCESS. A failure of r.stats was indistinguishable from a valid result, which is how it went unnoticed in the OSGeo4W CI in #7732.

Treating zero records as an error also removes the read from the zero length allocation.

I have not done the second improvement suggested in the issue, checking the child's exit status, because G_popen_close() discards it and that is a lib/gis change.

Reproducing it

Two maps of only NULL cells are enough, since r.coin runs r.stats with the flag that skips the no data value, so nothing is printed:

r.mapcalc "empty1 = null()"
r.mapcalc "empty2 = null()"
r.coin first=empty1 second=empty2 units=c

Before, that prints a full report with cat# 0 and all zeros, and exits 0. After, it exits 1 with ERROR: No data returned from r.stats.

Tests

test_r_coin_errors_without_data in the existing raster/r.coin/tests/test_coin.py does exactly the above and expects the module to fail. Checked against the unfixed build, where it fails with DID NOT RAISE CalledModuleError; the existing test_r_coin passes either way. clang-format and ruff report no changes.

I used an AI assistant while preparing this. I understand the change and can explain it.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.