stdout: changed some logging from stderr to stdout

[knutaros]

clear separation between trace logging to stdout and warning/error logging to stderr

[bolding]

I actually think the original use of stderr is correct. For two reasons:

1. stderr is for errors AND diagnostics
2. if stdout is used for diagnostics the concept of piping (in Linux) becomes difficult.

I know GOTM is not used in pipes - at least I'm not aware of it - but out of principle

[knutaros]

agreed. But then, could we change STDOUT in https://github.com/gotm-model/code/blob/master/src/fabm/gotm_fabm.F90#L1713 to LEVEL0 to write the reason (tracer values exceeding the valid range) for stopping the model to the same unit as the error message "Model state is invalid and repair is not allowed"? The other remaining output that might end up in stdout are some write(*,*) statements from the ode solvers, which could be changed as well such that gotm consistently writes to stderr only.

…

On 12/12/25 08:11, Karsten Bolding wrote: *bolding* left a comment (gotm-model/code#12) <#12 (comment)> I actually think the original use of stderr is correct. For two reasons: 1. stderr is for errors AND diagnostics 2. if stdout is used for diagnostics the concept of piping (in Linux) becomes difficult. I know GOTM is not used in pipes - at least I'm not aware of it - but out of principle — Reply to this email directly, view it on GitHub <#12 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AC2RV6XLUOZ6DFDUFAVAV4D4BJTDRAVCNFSM6AAAAACO2HKX3OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTMNBVGIYTMNRQHA>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[bolding]

Is the subroutine only used for reporting out of bounds?

It is clearly a diagnostic and should go to STDERR I think