feat: add dataset operations to sdkserver #869
Conversation
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
g-linville
requested review from
drpebcak,
thedadams,
njhale,
StrongMonkey,
iwilltry42 and
tylerslaton
pkg/gptscript/gptscript.go
Outdated
| if result.DatasetToolRepo == "" { | ||
| result.DatasetToolRepo = defaultDatasetToolRepo | ||
| } |
There was a problem hiding this comment.
I don't think this is needed given that you have a default set.
I do believe it is needed here, though, when running the SDK server in embedded mode:
gptscript/pkg/sdkserver/server.go
Line 154 in 45d444f
There was a problem hiding this comment.
Thanks. This is fixed
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
pkg/sdkserver/datasets.go
Outdated
| Name string `json:"dataset_name"` | ||
| Description string `json:"dataset_description"` |
There was a problem hiding this comment.
nit: you have camelCase and snake_case in your json tags. Can this be standardized?
There was a problem hiding this comment.
This went the opposite way I expected. Is there reason why you are doing snake_case? I think most of our stuff uses camelCase.
There was a problem hiding this comment.
Lol. I honestly have no idea. I'll go change it all to camelCase.
There was a problem hiding this comment.
Fixed. Sorry for the delay. Got stuck working on a test for another repo
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
Signed-off-by: Grant Linville <[email protected]>
This adds five new operations to the SDKServer to run the tools that we are using as the API for datasets. The SDKServer is basically just acting as a middleman between the client using the SDK, and the dataset API (the five tools in gptscript-ai/datasets#1).