Opened 3 years ago

Closed 3 years ago

#17261 closed defect (implemented)

Formalize our best-guess guard algorithm

Reported by: nickm Owned by: isis
Priority: Very High Milestone: Tor: 0.2.8.x-final
Component: Core Tor/Tor Version:
Severity: Normal Keywords: 028-triage, TorCoreTeam201510, tor-guard
Cc: isis Actual Points:
Parent ID: Points: medium
Reviewer: Sponsor: SponsorU

Description

We've made lots of progress in this area but final conclusions seem to elude us. Goal: Figure out what our very best idea is now, and write it down in a single document. Due: end of october.

Child Tickets

Change History (9)

comment:1 Changed 3 years ago by nickm

Keywords: TorCoreTeam201510 added

comment:2 Changed 3 years ago by isis

Cc: isis added
Keywords: tor-guard added
Severity: Normal

For the year 1 Sponsor U deliverables, we're supposed to update our specifications on the design of the current algorithm, and produce a white paper. Yet here it says:

Goal: Figure out what our very best idea is now, and write it down in a single document.

So… no whitepaper? And "our very best idea" means that we should specify the best design we can think of? Or specify how we think the current algorithm is likely to behave? (Or both?)

comment:3 Changed 3 years ago by nickm

The whitepaper is what we make it. I suggest that we do something in informal whitepaper standing, at the level of a technical proposal. Formalizing the current algorithm a little as a part of that proposal would rock.

"Our best idea" to me means the best design we can think of right now... should get written down in a concrete form so we can throw things at it.

comment:4 Changed 3 years ago by isis

Owner: set to isis
Priority: MediumVery High
Status: newassigned

comment:5 Changed 3 years ago by isis

Status: assignedneeds_review

I've documented the current algorithm (but not its failure modes, as #17262 mentions) in §5 of torspec.git/path-spec.txt, in my bug17261 branch.

Last edited 3 years ago by isis (previous) (diff)

comment:6 Changed 3 years ago by nickm

I don't see anything wrong with it, but I wasn't able to look super close. Still, that's an improvement over the big pile of nothing that's documented in path-spec.txt today. Soooo, merging!

We still need to write down our best guess as to what the path selection algorithm should be, of course.

comment:7 Changed 3 years ago by nickm

Status: needs_reviewnew

comment:8 in reply to:  6 Changed 3 years ago by isis

Status: newneeds_review

Replying to nickm:

I don't see anything wrong with it, but I wasn't able to look super close. Still, that's an improvement over the big pile of nothing that's documented in path-spec.txt today. Soooo, merging!

We still need to write down our best guess as to what the path selection algorithm should be, of course.


Okay, my "best guess" is up for review as a proposal in my bug17261-best-guess branch of my torspec repo.

comment:9 Changed 3 years ago by nickm

Resolution: implemented
Status: needs_reviewclosed

This is now proposal 259!

Note: See TracTickets for help on using tickets.