[jeckersb]

In #346 the manpage generation
was changed to mirror bootc, however it missed the important part that
bootc does not directly run cargo xtask but instead runs cargo run --release --package xtask -- ..., which is an important distinction
because the former requires the cargo-xtask binary to be present in
$PATH somewhere. In the RPM buildroot we don't have cargo-xtask so
this fails to generate manpages via packit (and would eventually FTBFS
in Fedora, too).

Closes: #355
Signed-off-by: John Eckersberg dev@eckersberg.com

[jeckersb]

well this partially fixes it, re: #355 (comment) still more to do

[cgwalters]

This is fine, but what was missed was copying https://github.com/bootc-dev/bootc/blob/main/.cargo/config.toml which I think is a better fix

[jeckersb]

This is fine, but what was missed was copying https://github.com/bootc-dev/bootc/blob/main/.cargo/config.toml which I think is a better fix

Yes, although that alias kinda obfuscates what is going on (hence my confusion)

[cgwalters]

Yes, although that alias kinda obfuscates what is going on (hence my confusion)

I agree it's subtle if one doesn't know about it, but it is documented in the first google search result (for me) for "cargo xtask" ➡️ https://github.com/matklad/cargo-xtask

So I think the right fix is to add that alias - and actually we should have a CI check that verifies that it works, I think that was the other real gap.

[jeckersb]

Yes, although that alias kinda obfuscates what is going on (hence my confusion)

I agree it's subtle if one doesn't know about it, but it is documented in the first google search result (for me) for "cargo xtask" ➡️ https://github.com/matklad/cargo-xtask

So I think the right fix is to add that alias - and actually we should have a CI check that verifies that it works, I think that was the other real gap.

Ah ok, I didn't realize that it's always just been an alias, instead assuming there was come cargo-xtask binary installed in the cargo path like all of the other cargo-XXX drop-ins. I'll change the first commit to add the alias.

[jeckersb]

Actually I went to add the alias but it's already there so something else is going on.

[cgwalters]

OK hum my offhand guess is this may be something with the Fedora RPM rust macros overwriting cargo.toml, IIRC I got burnt by that in the past, not sure why bootc doesn't hit the same thing.

[jeckersb]

OK hum my offhand guess is this may be something with the Fedora RPM rust macros overwriting cargo.toml, IIRC I got burnt by that in the past, not sure why bootc doesn't hit the same thing.

Yeah seems this is intentional:

- The `%cargo_prep` macro was adapted to unconditionally remove any pre-existing
  `.cargo` directory. In almost all cases, upstream settings in `.cargo/config`
  or `.cargo/config.toml` are incompatible with RPM build environments. This
  also removes a warning about potentially both of these files being present.