search

internetstandards / Internet.nl

Internet standards compliance test suite
https://internet.nl
164 stars 36 forks source link

Different score on Internet.nl vs. forked instance #1439

Closed apio-sys closed 2 weeks ago

apio-sys commented 2 weeks ago

I'm running a forked instance that will the basis for the French version to come here : https://fr.conformite-internet.fr/ . I'm up2date with the upstream main branch and there are no changes to the core (not any more since recent fixes), only customised code where it is supposed to be customised. Yet, I'm finding sometimes different test results between Internet.nl and my forked instance. For example, a site with not such a good score like this: https://fr.conformite-internet.fr/site/casabooking.com/400/ is getting 60% on my instance but on Internet.nl get's 47% here: https://internet.nl/site/casabooking.com/2843572/ . I have checked that there is no differences/custom scores in checks/scoring.py and otherwise the whole core is still the same. Can you see any reason for such a difference? Visually, the red flags are exactly on the same issues found.

Kind regards,

Joris.

mxsasha commented 2 weeks ago

Most likely: you're running the main branch, internet.nl is 1.8. RPKI has score impact in the main branch. I think that's in categories.py - see the scoring documentation for more details.

apio-sys commented 2 weeks ago

Thanks for your feedback @mxsasha . Indeed, I'm running off the main branch. So that explains the difference then. For future, is it better to keep running off the main branch or use a specific version?

mxsasha commented 2 weeks ago

Well, it depends. If you're moving closer to production use, better to follow a stable branch, as we are less likely to break things there. Also the lower rate of change makes rebasing easier.

apio-sys commented 2 weeks ago

OK many thanks for the explanation. I'll remain on main for the moment as otherwise I would miss out on all the recent improvements made for forked deployments (which are much appreciated).