build-commands: add: The build process fails when an individual entry…

[nandedamana]

… in the array fails.

Just making it clear so that one won't be hesitant to break down a long '&&'-ed one-liner.

[smcv]

Thanks, it would be good to document this behaviour.

The first line of the commit message is rather long (and Github is giving you a hint that this might be a problem by ellipsizing it). Maybe something like this?

flatpak-manifest(5): If one command fails, the whole build fails

Documenting this makes it more obvious that it's valid to write

    - foo
    - bar
    - baz

and unnecessary to combine those commands into
`foo && bar && baz` to get sensible error behaviour.

[smcv]

Just making it clear so that one won't be hesitant to break down a long '&&'-ed one-liner.

This is good reasoning that ought to be captured in the commit message for the benefit of future maintainers. (For example see https://cbea.ms/git-commit/)

[smcv]

Please squash your two commits into one, ideally with an explanatory commit message like the one I suggested in #601 (review)

[nandedamana]

Just making it clear so that one won't be hesitant to break down a long '&&'-ed one-liner.

This is good reasoning that ought to be captured in the commit message for the benefit of future maintainers. (For example see https://cbea.ms/git-commit/)

Yes, thanks. I was in a hurry and was just copy-pasting something relevant from the change itself.

BTW, this commit originated from a similar hesitation that I had and a YAML pitfall that made me question the behaviour of build-commands. It is documented here: https://nandakumar.org/blog/2024/05/yaml-boolean-trap.html

[smcv]

Yeah, it's clearly a bug that flatpak-builder doesn't run false as a command in that situation, but also doesn't error out with some semi-useful message like "expected string, but found boolean". Please report that as an issue if nobody has done so already - and I'm sure a PR to solve it would be appreciated, if you have some time.

false is not even the worst YAML trap: if you're discussing ISO country codes, it's easy to get tripped up by Norway being no, which (if unquoted) also gets interpreted as boolean FALSE ("the Norway problem").

[nandedamana]

Please squash your two commits into one, ideally with an explanatory commit message like the one I suggested in #601 (review)

Your commit message looked fine, so I just reused it.

[nandedamana]

@smcv Anything else to be done from my part?