search

Derpseh / Spyglass

Nationstates Sheet-generator for easily finding approximate update times
GNU General Public License v3.0
11 stars 10 forks source link

Spyglass Times are Wrong #38

Closed EWSouls closed 1 year ago

EWSouls commented 1 year ago

Something is fucky with spyglass times, and it seems to come from the 3.0 refactor. All large regions show incorrect time stamps, as tested on multiple OS' (at least Windows and Linux) and across the past several days. The following information comes from today's (4/6) dump, but I also personally checked and found the issue in a 3/31 sheet from the day 3.0 released, and Durm observed it 4 days ago as well.

These are sheets from 2.0.2 and 3.0.0 spyglass respectively, utilizing the same dump file:

2.0.2 - https://i.imgur.com/7npFoPi.png 3.0.0 - https://i.imgur.com/nHN2inD.png

You can see in the former, the expect sequence occurs: 0:1:15, 0:1:16 [large region], 0:1:25 [subsequent to large region] Meanwhile in the latter, it goes 0:0:40, 0:1:2 [large region], time travelling back to 0:0:51 [subsequent to the large region]

The about 10-11s gap to the subsequent region is correct, but the time stamp on the large region is drastically incorrect.

I also checked the dump regions.xml in notepad++ myself to verify this. In the same dump file that Spyglass ran off of, the timestamps for those 3 example regions are as follows:

Republic of Cuba: LASTUPDATE 1680753687 LASTMAJORUPDATE 1680753687 LASTMINORUPDATE 1680710432

1680753687 = GMT: Thursday, April 6, 2023 4:01:27 AM

United States of America: LASTUPDATE 1680753688 LASTMAJORUPDATE 1680753688 LASTMINORUPDATE 1680710432

1680753688 = GMT: Thursday, April 6, 2023 4:01:28 AM

Refrigerator: LASTUPDATE 1680753697 LASTMAJORUPDATE1680753697 LASTMINORUPDATE1680710435

1680753697 = GMT: Thursday, April 6, 2023 4:01:37 AM

Two things stand out to me here: 1) Nothing is fucky with these timestamps. They follow an expected pattern of 0:1:27, 0:1:28 [large region], 0:1:37 [subsequent to the large region], which is what we see in the 2.0.2 sheet. 2) My understanding was the for Major, Spyglass reports "lastupdate" and not "interpolated" - if that's the case, it's extra odd that these times do not match either sheet, and are about 10s behind the 2.0.2 sheet, while being about 20s ahead of the 3.0.0 sheet.

If I am incorrect and Major times are also interpolated, then I suppose it's clear some interpolation logic likely got fucky in the 3.0 refactor. If Major times are supposed to be lastupdate, then...wtf?

EWSouls commented 1 year ago

Narrowed down to an issue with CumuNation logic. Aav is testing a fix, to be pulled into main imminently.

EWSouls commented 1 year ago

Reopening to make Aav happy.