equinor
smspec_node is_historical has false positives
Currently only checks if vector ends with H.
Several examples of vectors which return true, but should return false, exist,. e.g.: MONTH and TCPUH.
History is defined through keywords WCONHIST and WCONINJH, hence vectors that are not of ecl_smspec_var_type: ECL_SMSPEC_WELL_VAR, ECL_SMSPEC_GROUP_VAR or ECL_SMSPEC_FIELD_VAR should not be possible as far as I know.
Checking vector ecl_smspec_var_type should therefore reduce the risk of false positives.
Is this still relevant? Looks a lot like https://github.com/equinor/ecl/pull/707 . It'd be nice to have documentation of how to detect historical data.
Cool. Does ECLIPSE have any documentation of what constitutes historical data? Or is this a naming convention and we just have to guess?
I haven't seen a clear definition, on the other hand: I must admit I haven't read the whole thing 😬🙈 But as I mention above, as far as I know, only field, group and well vectors can be historical (potentially also completions, but I don't think so). And the keyword part will always end on an H like the current implementation (afaik). So my suggestion is to reduce risk of false positives by only checking the types above. But I would love for someone with years of experience with Eclipse vectors to verify that they haven't seen historical vectors of other types before we conclude on the matter 😅
My suggested (maybe not completely a) solution should of course be easy to implement.