Does Globus default to partial string matching? #126
Description
In esgpull, the query parameter of the API is used, so that we can leverage both inclusion and exclusion.
We've noticed that with the bridge API, the matching seems to be done with partial strings, rather than exactly. This differs from previous Solr behaviour.
Solr example:
$ esgpull config api.index_node --default
[api]
index_node = "esgf-node.ipsl.upmc.fr"
$ uv run esgpull search experiment_id:amip-hist --hints experiment_id
[
{
"experiment_id": {
"amip-hist": 2522
}
}
]
Using the bridge:
$ esgpull config api.index_node esgf-node.ornl.gov/esgf-1-5-bridge
[api]
index_node = "esgf-node.ornl.gov/esgf-1-5-bridge"
$ esgpull search experiment_id:amip-hist --hints experiment_id
[
{
"experiment_id": {
"amip": 108205,
"hist-nat": 94873,
"hist-GHG": 89217,
"hist-aer": 81048,
"hist-sol": 67465,
"hist-volc": 57438,
"esm-hist": 56232,
"hist-noLu": 26673,
"hist-stratO3": 26218,
"hist-piAer": 20015,
"hist-totalO3": 19649,
"hist-piNTCF": 19161,
"amip-hist": 18933,
"hist-1950HC": 14917,
"hist-1950": 14312,
"land-hist": 12495,
"hist-CO2": 11102,
"hist-bgc": 9405,
"amip-p4K": 8467,
"amip-4xCO2": 7464,
"amip-future4K": 5713,
"amip-piForcing": 5545,
"amip-lfmip-pdLC": 5267,
"amip-lfmip-rmLC": 5197,
"amip-m4K": 4763,
"hist-resIPO": 3269,
"amip-p4K-lwoff": 3136,
"hist-spAer-all": 3125,
"amip-lwoff": 3071,
"land-hist-cruNcep": 2424,
"hist-resAMO": 2271,
"hist-GHG-cmip5": 2250,
"hist-nat-cmip5": 2250,
"hist-aer-cmip5": 2130,
"land-hist-princeton": 2034,
"land-hist-altStartYear": 1286,
"amip-a4SST-4xCO2": 1124,
"land-hist-wfdei": 974,
"amip-hld": 838,
"amip-lfmip-pObs": 492,
"amip-TIP": 360,
"amip-TIP-nosh": 351,
"land-hist-altLu1": 227,
"land-hist-altLu2": 227,
"hist-lu": 9
}
}
]
I also noticed we can force an exact match with quotes around the value, so I will use this fix on esgpull for now:
$ esgpull search experiment_id:'"amip-hist"' --hints experiment_id
[
{
"experiment_id": {
"amip-hist": 18933
}
}
]
I don't know if this is to be fixed on your side, it should probably be documented it if kept like this.