Feedback

[ValueRaider]

I am not too satisfied with the way yahoo_cache is operating ranaroussi/yfinance#1665 (comment)

@rickturner2001 would you mind elaborating? I'm receptive to criticisms and suggestions.

[rickturner2001]

I believe you've done an exceptional job. What I intended to convey is that the integration of this might not align seamlessly with an existing yfinance project. To illustrate, if I were to incorporate caching into a project that utilizes yfinance, my anticipation would be to execute something along these lines:

import some_package from some_lib
import yfinance as yf

# Cached session instance
config = {...}
session = CachedSession(config=config)

# code ...
yf.download(..., session=)

In essence, the only task I would need to undertake here would be generating the session instance, after which the program would maintain its usual functionality.

In summary, my concern pertains to the manner in which the API harmonizes with yfinance, rather than any issues with the implementations themselves.

[rickturner2001]

I understand the concept, but I'm a bit puzzled as to why caching the URL isn't a feasible approach here. How can you determine whether something being retrieved has been previously fetched, enabling you to potentially provide the cached outcome instead of initiating a redundant fetch?

Regardin Python's GIL limitations, sure, we cannot really achieve true parallelism, but since we're talking about I/O-bound operations, Multi-threading in Python still allows for concurrency. Woulnd't you see that as a plausible option?

[ValueRaider]

but I'm a bit puzzled as to why caching the URL isn't a feasible approach here

How to handle stock splits, and dividends if you use adjusted prices? Is feasible if you're OK wiping the cache and re-fetching everything. Then how do you update current active trading interval?

How can you determine whether something being retrieved has been previously fetched

... it's in the cache. A better question would be: when does (some of) cache data need to be refetched?

talking about I/O-bound operations, Multi-threading in Python still allows for concurrency. Woulnd't you see that as a plausible option?

It's a cost-benefit tradeoff. Benefit = what you describe, maybe, in some scenarios. Cost = development time.

[ValueRaider]

Does this improved README explain sufficiently how cache works? #38

[rickturner2001]

I see. Yes, the new README clearly points out the cababilities and limitations of the API

[ValueRaider]

Given your interest in download, I've implemented it in #41. Actually wasn't difficult, but not stress-tested it yet.