add support for the third secondary backoff

[DaveHickeyDev]

from github docs:
If you exceed a secondary rate limit, you will receive a 403 or 429 response and an error message that indicates that you exceeded a secondary rate limit. If the retry-after response header is present, you should not retry your request until after that many seconds has elapsed. If the x-ratelimit-remaining header is 0, you should not retry your request until after the time, in UTC epoch seconds, specified by the x-ratelimit-reset header.

Otherwise, wait for at least one minute before retrying. If your request continues to fail due to a secondary rate limit, wait for an exponentially increasing amount of time between retries, and throw an error after a specific number of retries.

Would be cool if this library implements the "otherwise" part, wait for an exponentially increasing amount of time between retries, and throw an error after a specific number of retries.

[gofri]

Hi @DaveHickeyDev,
thanks for contributing your idea!

I don't mind implementing that in principle (or at least approving a PR for that),
but I haven't encountered such issues in reality and I prefer to avoid adding functionality if it has no practical use.

Have you ever encountered the issue?

p.s. note there's already a comment in the code that describes this decision @ detect.go:

func parseSecondaryLimitTime(resp *http.Response) *time.Time {
	if !isSecondaryRateLimit(resp) {
		return nil
	}

	if sleepUntil := parseRetryAfter(resp.Header); sleepUntil != nil {
		return sleepUntil
	}

	if sleepUntil := parseXRateLimitReset(resp); sleepUntil != nil {
		return sleepUntil
	}

	// XXX: per GitHub API docs, we should default to a 60 seconds sleep duration in case the header is missing,
	//		with an exponential backoff mechanism.
	//		we may want to implement this in the future (with configurable limits),
	//		but let's avoid it while there are no known cases of missing headers.
	return nil
}

[amenocal]

👋🏼 So I actually ran into 429 error message where it seems that the github_ratelimit didn't pick up.

This is the error message my utility logged:

adding repository to team:  product-security repo pull Unknown error PUT https://api.github.com/orgs/<org>>teams/product-security/repos/<org>>/<repo>: 429  [] PUT https://api.github.com/orgs/<org>/teams/product-security/repos/<org->/<repo>: 429  []

and you can see here how my cli tool is creating the client and here is the API call where I got the 429 from 😅

Granted I'm still in v1.1.0 and I'm using also go-github's default primary ratelimit handler, so I'm not sure if this two would interfere with one another