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
Thanks for this great tool, this is really valuable !
I'm currently migrating to the V5 API and was working about integrating prices.
This json is big (around 400Mb on disk) and it's a cumbersome to process it because it takes a lot of RAM just to load the json in memory.
Have you considered to do an export in the json lines format (jsonl): http://jsonlines.org/ ?
The format is one json object per line, this way you can process the files by iterating lines by lines without having to load everything in memory, like a stream.
This could also applied to other exports.
The text was updated successfully, but these errors were encountered:
I looked into jsonlines, but it doesn't seem to be a standardized tool. Is this really the best approach to helping folks who can't load in the full data set at once? Is there a feature of your language that will allow for lazy loading you might be able to take advantage of?
It also looks like JSONLines wants constant structs, but that would mean the loss of several data points, like meta data. If you have suggestions, I'm all ears!
I agree that's more a convention than a real standard, but it doesn't require to reinvent the wheel since it's 99% based on json.
For the metadata, it's not forbidden to put as the first line, or you can imagine to provide it as a separate file.
On my side, I had to preprocess the file using the streaming API of jq (https://stedolan.github.io/jq/manual/#Streaming), it required more work but it worked.
Thanks for this great tool, this is really valuable !
I'm currently migrating to the V5 API and was working about integrating prices.
This json is big (around 400Mb on disk) and it's a cumbersome to process it because it takes a lot of RAM just to load the json in memory.
Have you considered to do an export in the json lines format (jsonl): http://jsonlines.org/ ?
The format is one json object per line, this way you can process the files by iterating lines by lines without having to load everything in memory, like a stream.
This could also applied to other exports.
The text was updated successfully, but these errors were encountered: