google / self-published-geo Goto Github PK

License: Creative Commons Attribution 4.0 International

Python 100.00%

self-published-geo's Introduction

draft-self-published-geo

IETF document for experimental means whereby IP address owners/maintainers can self-publish geolocation information for interested consumers.

This document has been published as RFC 8805.

self-published-geo's People

Contributors

Stargazers

Watchers

Forkers

ekline wkumari neotim isabella232 ioggstream ghas-results

self-published-geo's Issues

RFC 4180 cited as normative but not actually respected

RFC 4180 (CSV) mandates things like fixed number of commas, whereas the document current is loose about this (says it's optional).

Possible Resolutions:

[1] make commas mandatory, or
[2] make 4180 informative.

Given the existing deployed formats and parsers we should probably choose option 2.

Consider renaming "country" field

Possible alternatives include "economy", "ISO3166.1a2 field", et cetera.

much (most) of section 7 should be reevaluated

Section 7.3, in particular, was never really supported.

Much of section 7 should be moved to an appendix or excised entirely.

security section should discuss claims of unowned IP space

"""
b) The security considerations section should specifically note the
possibility of an attack where ISP Bob claims IP space held by ISP
Alice.
"""

address all idnits

See http://www.ietf.org/tools/idnits/ .

We should do this once all all other issues are closed and changes to the text are essentially done.

alpha2iso should allow non-conforming (EU) codes

Minor nit, can you modify the paragraph about the semantic intent and conformance for alpha2iso to permit use of EU? It's in the reserved/exceptional set, arguably this is "ok" as-is and you can close the file. https://www.iso.org/obp/ui/#iso:code:3166:EU puts it "off to one side" of the official countryCode list.

A nit-picker might disagree. But equally, not. I'm fine if you choose to close this with a sigh.

minor nit: can you revise the code example

865c865
<       if not self._IsCountryCode2Correct(fields[1]):
---
>       if not self._IsAlpha2CodeCorrect(fields[1]):
907,908c907,908
<   def _IsCountryCode2Correct(self, country_code_2):
<     if len(country_code_2) == 0:
---
>   def _IsAlpha2CodeCorrect(self, alpha_2_code):
>     if len(alpha_2_code) == 0:
910c910
<     if len(country_code_2) != 2 or not country_code_2.isalpha():
---
>     if len(alpha_2_code) != 2 or not alpha_2_code.isalpha():
924c924
<     if not self._IsCountryCode2Correct(parts[0]):
---
>     if not self._IsAlpha2CodeCorrect(parts[0]):

reclassify as Informational

The original classification (from back in ~2012) was Experimental. But given that we're documenting a format already in use (since before the original document was posted), Informational is probably a better classification.

extract python code and make sure unittests pass

Additionally, consider:

[1] python2 versus python3?

[2] see if CI can be set up (though this code requires the IP address library)