sgmenda / browser-fingerprinting Goto Github PK

I had a chat (on github) with @pes10k about this, that there are only so many timezone aliases - you can use this PoC - I can add you to that private repo if you like, just let me know

I designed it to see how many unique results I could get using timezone offsets, and I'm using date.parse() and epoch time as it's very difficult for extensions to tamper with, not the PoC, but the actual test on tzp, As a side effect, it groups aliased or identical timezones together

Now the minimum test for max results is 5 years (2 dates per year). This gives 374 timezone groups and means all dates, calendars etc would not be "broken" (i.e unintended results for the end user) for any date ever.

I confused myself in the chat, it had been quite a while since I built the PoC, but essentially, if you were to only use years going forward, say from 2018, just use 2018, 2019, 2020, 2021 in the text field (so some calendar backward compat), then you could get away with as few 70 timezones, of which 24 will be unique (i.e no randomizing possible, and those are likely to always be a problem), but other 46 groups covering the remaining 431 timezones allow randomizing with no "breakage" - i.e, as long as it is made clear in documentation that it is forward compatible only.

Note: some of those groups are small (2 or 3), but I doubt you can do anything about them. And some can be misleading, for example 2018+ Antarctica/McMurdo, Pacific/Auckland, Pacific/Fiji .. or my 5 year PoC example: Antarctica/McMurdo, Pacific/Auckland ... who is at McMurdo, might as well just say you're New Zealand time?

edit: this was meant to be about your timezone comments at https://github.com/sgmenda/ideas