variable name philosophy

[mvhulten]

Concerning naming, consistency between README and scripts and updating, is there a philosophy for the shell variables?

Some examples:

• In control_tsmp2.sh there is $expid, gotten from either $EXP_ID further up (where user can adjust it), but in README.md there is $sim_id
• There is CASE_ID and caseid
• …

One sort of principle is visible, of course: user defines XXX_YYY and script picks it then up as xxxyyy, but other things work different again. Also there is npnode=$npnode_u and similar, and so on…

The actual question is rather: if I find it messy do I not understand properly (then we should talk), or is it how I think it is and should it be like that, or should I try to make it more consistent?

@s-poll, apropos, in the TSMP2 repo (at least that stages-2024 branch), in fact, lower-case hostnames are used (so they are not up-cased with ^^). Someone like-minded must've changed it thus. :-) (in retrospect, of very little importance, and I'll only touch it if I must touch the line for another reason as well)

[mvhulten]

A bit more specific to this issue (an exaple):

Say, I compile eCLM only (so ./build_tsmp2.sh --eCLM—if that would work überhaupt, but see in the relevant TSMP pull request for that—oh and here are some capital letters whereas in the wfe they are all lower-case (which I prefer, but more important is that they will be the same)) and then when I run it, by default it does coupled becasue by default

MODEL_ID=ICON-eCLM-ParFlow

[kvrigor]

I think the "right" answer here depends on which convention we adopt. Perhaps the Google shell styleguide would be a good start.

[s-poll]

Just in short, I will add more information at a later stage.

The variable name philosophy is that variables which ends with _u are user-defined variables and are not used by the system itself. These are transferred to variables without _u-ending. The difference between EXP_ID (input) and the expid (wfe internal usage) are that the exp_id is a modified (lower-casing, removing of separator) version of EXP_ID, which is connected with the separation in the directory names.

I admit that the naming scheme might not always be intuitive. I am open for suggestions.

[mvhulten]

The suggestion is here, first as a separate feature branch, but it took too long and was afraid of merging conflicts arrising, so I merged it already to my main improvement branch that can be merged (if you have no more issues).