Throttle and set_rate_limit have limited usability

[jaraco]

Throttle and set_rate_limit have limited usability

[jaraco]

Attached a patch anyway. Please test against all environments as I only tested against python 2.7 on linux.

Bottom line:

• It now works as I think it was intended
• Makes two calls to time.time() per function call but I don't think there's gonna be a performance issue.
• Can't see how this could break existing code using set_rate_limit. Still, as was also the case in the previous implementation, one should be wary of where to start the rate limiting, as putting it too early can slow the client's initial connect.
• Testing takes 2 more seconds to run, one of which is a time.sleep()

---

Original comment by: Nick Raptis

[jaraco]

Would you consider uploading the patch as a pull request, either here in bitbucket or on the github mirror? That'll help me review and assign attribution.

---

Original comment by: Jason R. Coombs

[jaraco]

I was trying to avoid that actually.
I have no idea how to properly manage branches in mercurial and how bitbucket does pull requests, but it's as good time as any to try it.
Just tell me, should I open a new branch for my commit or is it all the same to you and default is ok?

---

Original comment by: Nick Raptis

[jaraco]

It's easy. Just 'fork' the repo, commit your changes in a clone (of either repo) (or edit the file in your fork directly through the web), push the changes to your fork (if necessary), and finally, click Pull Request in your fork.

You may use the default branch for all of this.

---

Original comment by: Jason R. Coombs

[jaraco]

Merged in nickraptis/irc (pull request #18)

Better rate limiting (fixes issue #25)

→ <<cset 441b50f97e73>>

---

Original comment by: Jason R. Coombs

[jaraco]

Better rate limiting (fixes issue #25)

Previous reate limiting implementation did not work when bursts of
calls were separated by a significant amount of time.
Now limits based on the time of last call instead of the time
the Throttler was started.

→ <>

---

Original comment by: Jason R. Coombs