You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on May 18, 2024. It is now read-only.
Sometimes we would like to use input versions for formats which are currently not supported by finch.
The only option we then currently have is to provide the data as the source, from which we can only have one per input.
However, it should be possible to compare any dataformats, even ones which are not built into finch.
For example, let's assume we want to compare how the runtime is being affected when we use GRIB, compared to loading the data from some database (i.e. FDB).
Since we currently cannot write new GRIB versions, we cannot compare the two data formats.
We can solve this problem by adding a new CUSTOM data format, which can be added with Input.add_version().
When adding the new version, a function must be provided which defines how the data is retrieved.
We could even generalize this concept such that all input formats have a dedicated data retrieval function.
The text was updated successfully, but these errors were encountered:
Currently, we can provide multiple "unknown" input versions via a single source, because the requested version gets passed as an argument.
So we could actually compare GRIB with FDB by branching in the source between the two input versions, depending on some condition derived from the requested version object.
However, this functionality still provides its limits, since the current version class doesn't support a lot of configurable options as of now.
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Labels
coreIssues adressing the finch library itselfenhancementNew feature or request
Sometimes we would like to use input versions for formats which are currently not supported by finch.
The only option we then currently have is to provide the data as the source, from which we can only have one per input.
However, it should be possible to compare any dataformats, even ones which are not built into finch.
For example, let's assume we want to compare how the runtime is being affected when we use GRIB, compared to loading the data from some database (i.e. FDB).
Since we currently cannot write new GRIB versions, we cannot compare the two data formats.
We can solve this problem by adding a new
CUSTOM
data format, which can be added withInput.add_version()
.When adding the new version, a function must be provided which defines how the data is retrieved.
We could even generalize this concept such that all input formats have a dedicated data retrieval function.
The text was updated successfully, but these errors were encountered: