the opam lock command exits with a non-zero exit code when dependencies are not yet installed in the current switch

[progressive-galib]

fixed #5520

[kit-ty-kate]

thanks for your contribution. Aside from the two individual comments, i can note that the if clause should change from as its return value isn't used in the true case anymore. Could you change the shape of the if to the following?:

if ... then
   error_and_exit... ;
OpamPackage.Set.add nv acc

[kit-ty-kate]

The error message doesn't correspond to the action anymore (skipping vs. aborting)

[progressive-galib]

done but my new error msg might not make any sense grammatically,

as English is not first language help a bit, pls.

[progressive-galib]

my new error msg might not make any sense grammatically,

as English is not first language help a bit, pls.

[progressive-galib]

@kit-ty-kate done merge please

also can i talk to you about something else here or if you could msg me somewhere else ? i want to contribute more to opam more, i am just getting started so if you could suggest some easy/small issues and walk me through the source tree/code structure that would be nice

[progressive-galib]

@rjbou @kit-ty-kate any update ?

[kit-ty-kate]

Hi, sorry for the wait.
First of all, while i know English isn't your first language, i'd like to invite you to use a less pushy language when requesting action from us maintainers here and elsewhere, even more when requesting action during the winter break when the change isn't critical.

Regarding your request for discussion, both @rjbou and i have our email addresses displayed on our github profile, feel free to send us an email.

Regarding this here PR, i don't think the new behaviour matches the requested behaviour in #5520. I'm not sure it makes sense to stop as soon as a missing dependency is detected. Maybe keeping the same error message as before but simply returning the fact that an error occurred from this function, then use it to call OpamStd.Sys.exit_because if there was an error at the end of OpamCommands.lock.lock might be a better behaviour. This way the behaviour for (non-script) users of opam lock does not change while allowing script users to make sure the dependencies are all installed.
What do you think? As a user, how would you expect opam lock to work?