-
Notifications
You must be signed in to change notification settings - Fork 15
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Feature]: Have NCO detect all errors on first run #535
Comments
One potential solution is to use the fault tolerant variable list, suggested by @czender Or maybe @czender can advice if there are alternatives here. |
Yes the |
I am coming up to speed on what you are discussing. After reading #533, my impression is that timeseries generation is working as desired (i.e., failing quickly when non-present variables are requested), and that climo generation is not because it does not fail quickly enough when non-present variables are requested. Is this the essence of the problem? If not, please elaborate. Also, why are these tasks being invoked with non-present variable names in the first place? Once I understand both of these issues, I can be more helpful :) |
Yes, that's correct.
|
OK. Say |
I'm thinking it would process all available variables and then fail listing "notpresent1, notpresent2 not found". Currently, it would fail with "notpresent1 not found" and the user would have no idea that
I believe the reason |
It is possible to implement this desired behavior with a change to Note that comparing the requested |
I think for monthly climatology, it is not a bad idea to just include all variables (as default), especially for v3, the output variables are much reduced. |
Alright, we can do that as a default. But what do we want the behavior/implementation to be when users specifically set |
I leave the desired behavior up to you two. FWIW, my "philosophy" is that it's more dangerous to fix user input errors than it is to just fail so that users are forced to learn what's in their datasets. Otherwise they'll specify non-present variables and then be surprised when those variables to not appear in successful climatologies. I think we all want to avoid that situation, that Ryan called silently ignoring bad input. This is a grey area, though, in that I think Ryan's original point was that it's currently too difficult for users to learn what's in the dataset. So I agree that some form of automated |
Good question, if a user wants to get the specific set of |
I agree the logic to check what variables zppy hands ncclimo should live in zppy. |
@czender What would be the command to do this? If that's something
I agree. That's why I think the command should still fail, but with a list of all failures instead of just the first one encountered.
Yes, I think that's fine. I just need to know how to check that without actually attempting to generate the climatologies first. (See first point in this post). |
When users do not provide an explicit Examples:
|
Isn't zppy written in python? Just open the file with python netcdf lib and query for the variables. Don't rely on external programs from NCO. |
I couldn't agree more. This is something for |
Just remember that a valid |
How will this affect the next version number?
Small improvement (increment PATCH version)
Is your feature request related to a problem?
When running the
climo
task on multiple variables, only the first error is reported. I.e., in #533, it took 16 runs to find out 14 variables weren't supported onclimo
whereas it only took 1 run to find out 10 variables weren't supported onts
.Describe the solution you'd like
The
climo
task should report all errors in variables rather than simply stopping after encountering one error.Describe alternatives you've considered
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: