Skip to content

GitLab

Switch branch/tag
  1. 30 Aug, 2021 1 commit
  3. 22 Aug, 2021 1 commit
  5. 01 Apr, 2021 1 commit
  7. 12 Nov, 2020 1 commit
  9. 08 Jun, 2020 1 commit
  11. 02 May, 2020 1 commit
  13. 27 Apr, 2020 1 commit
  15. 21 Apr, 2020 1 commit
    • Steven Allen's avatar
      fix: minimize time holding wantlist lock (#361) · 9fc4a368
      Steven Allen authored 
      * fix: minimize time holding wantlist lock

Instead of holding the lock the entire time we prepare a message, hold the lock
while we retrieve the wantlist entries, process the entries without the lock,
retake the lock, then mark entries as sent.

This means:

1. We never sort entries while holding the lock.
2. We allocate exactly three times while holding the lock (once per entry list).

* fix: address code review
      9fc4a368
  17. 23 Mar, 2020 2 commits
  19. 16 Mar, 2020 1 commit
  21. 13 Mar, 2020 1 commit
  23. 10 Mar, 2020 1 commit
  25. 06 Mar, 2020 2 commits
  27. 02 Mar, 2020 1 commit
  29. 17 Feb, 2020 2 commits
  31. 13 Feb, 2020 1 commit
    • Steven Allen's avatar
      feat: debounce wants manually · 777c0d9a
      Steven Allen authored 
      This:

* Makes it easy to send immediately if we wait too long and/or if we have enough
to send.
* Is significantly more efficient than the debounce library as it doesn't spin
off a bunch of "after" timers.

fixes #245
      777c0d9a
  33. 12 Feb, 2020 1 commit
  35. 10 Feb, 2020 1 commit
  37. 30 Jan, 2020 2 commits
  39. 16 Dec, 2019 1 commit
  41. 09 Dec, 2019 1 commit
  43. 05 Dec, 2019 2 commits
  45. 15 Oct, 2019 1 commit
  47. 02 Oct, 2019 1 commit
  49. 01 Oct, 2019 1 commit
  51. 24 Sep, 2019 4 commits
  53. 07 Sep, 2019 1 commit
    • Steven Allen's avatar
      engine: tag peers based on usefulness · 9d580a65
      Steven Allen authored 
      This patch tracks two usefulness metrics: short-term usefulness and long-term
usefulness. Short-term usefulness is sampled frequently and highly weights new
observations. Long-term usefulness is sampled less frequently and highly weights
on long-term trends.

In practice, we do this by keeping two EWMAs. If we see an interaction within
the sampling period, we record the score, otherwise, we record a 0. The
short-term one has a high alpha and is sampled every shortTerm period. The
long-term one has a low alpha and is sampled every longTermRatio*shortTerm
period.

To calculate the final score, we sum the short-term and long-term scores then
adjust it ±25% based on our debt ratio. Peers that have historically been more
useful to us than we are to them get the highest score.
      9d580a65
  55. 14 Aug, 2019 1 commit
  57. 03 Jul, 2019 1 commit
  59. 02 Jul, 2019 1 commit
  61. 14 Jun, 2019 1 commit
  63. 12 Jun, 2019 1 commit