[tor-bugs] #23391 [Metrics/Metrics website]: Add changes section to Tor bridge descriptors specification
Tor Bug Tracker & Wiki
blackhole at torproject.org
Mon Sep 4 09:35:14 UTC 2017
#23391: Add changes section to Tor bridge descriptors specification
-------------------------------------+------------------------------
Reporter: karsten | Owner: metrics-team
Type: enhancement | Status: needs_review
Priority: Medium | Milestone:
Component: Metrics/Metrics website | Version:
Severity: Normal | Resolution:
Keywords: | Actual Points:
Parent ID: | Points:
Reviewer: | Sponsor:
-------------------------------------+------------------------------
Comment (by karsten):
Replying to [comment:3 iwakeh]:
> Yes, it is important to have such a change description/listing.
>
> To have a more complete changelog more items should be included. From
the simple
[https://gitweb.torproject.org/collector.git/log/?qt=grep&q=sani git
search]:
Ah, I guess we'll first have to decide what should go into this changes
section. My plan was to only include changes to the specification, that
is, changes to the way how we're modifying original bridge descriptors
when sanitizing them. Now, some of the changes you suggest don't affect
the specification but only the implementation. Let me explain using the
first example:
> * Retain padding-counts lines when sanitizing descriptors.
Karsten Loesing 2017-05-11
These "padding-counts" lines never showed up in bridge descriptors until a
few days before this commit. And when they first showed up, CollecTor
rejected the entire descriptor, printed out a warning, and continued with
the next descriptor. After we checked that these lines can go in without
change, this commit simply made sure they're copied over.
In summary, whenever an original bridge descriptors did not contain such a
line, the sanitized descriptor didn't either. And if the original
descriptor contained this line, the sanitized one did as well. That's also
why we didn't bump the version number. There was no change to the
specification.
This is different for changes where we're modifying line contents, adding
lines that didn't exist before, truncating lines, or even removing lines
entirely. Those are all changes that would require a version number bump
and probably an entry in this new changes section.
Maybe think of it in a different way: If our implementation would default
to accepting unknown lines and only touch lines that we modify in the
sanitizing process, these new lines in original descriptors would not even
produce a Git commit. I'm not saying it's a better approach, it's just an
attempt to distinguish the two types of changes.
> * Retain hidserv-* lines when sanitizing descriptors. Karsten
Loesing 2016-11-23
Same.
> * Sanitize TCP ports in bridge descriptors. Karsten Loesing
2016-09-18
This one is in my list, too.
> * Sanitize bridge descriptors with ed25519 lines. Karsten Loesing
2015-06-13
In my list, too.
> * Parse "published" lines from bridge statuses. Karsten Loesing
2014-10-13
This one is tricky. We did write such a line ourselves before, and with
this change we're copying that line from the original descriptor. I guess
we should add it, even though it didn't lead to a version number bump.
> * Clarify ntor-onion-key lines and @type versions. Karsten
Loesing 2014-06-10
That's just a documentation patch.
> * ...
>
> CollecTor's
[https://gitweb.torproject.org/collector.git/tree/CHANGELOG.md changelog]
is also a source for change info.
> Maybe, there's a way to automate the list? All bridge desc sanitation
related changes in a release tag without test related commits ...
> Well, at least a source for adding historical changes manually.
Maybe we can use this list as source for discussion what should go into
changes and what should not? Some of these changes are tricky and require
per-case inspection and possibly discussion.
Would you want to create such a list?
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23391#comment:4>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
More information about the tor-bugs
mailing list