regression: mismatched types (due to stabilization of `PathBuf::add_extension`)

[BoxyUwU]

• https://crater-reports.s3.amazonaws.com/beta-1.91-3/beta-2025-09-21/gh/mbhall88.drprg/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.91-3/beta-2025-09-21/reg/drprg-0.1.1/log.txt

[INFO] [stdout]    --> src/builder.rs:265:9
[INFO] [stdout]     |
[INFO] [stdout] 263 |     fn prg_index_path(&self) -> PathBuf {
[INFO] [stdout]     |                                 ------- expected `PathBuf` because of return type
[INFO] [stdout] 264 |         let ext = format!("k{}.w{}.idx", self.pandora_k, self.pandora_w);
[INFO] [stdout] 265 |         self.prg_path().add_extension(ext.as_str().as_ref())
[INFO] [stdout]     |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ expected `PathBuf`, found `bool`

[fmease]

Now shadowed by the recently stabilized add_extension (since unstable methods don't shadow stable ones). So this has to be due to #145209 (not cratered).

[fmease]

I don't remember what T-libs-api/T-libs's stance is on such kind of breakages? Are they generally accepted nowadays unless the breakage is enormous / widespread?

[ChrisDenton]

This is "allowable breakage" but, yeah, it's ultimately up to libs-api to decide. At a minimum affected crates should be contacted and patched.

[fmease]

it's ultimately up to libs-api to decide.

cc @joshtriplett (#127292 (comment))

At a minimum affected crates should be contacted and patched.

cc @hanna-kruppe @tgross35 (#145209)