[OPEN] Map suddenly stops working

Expected behavior
When I attempt to start an unchanged map that worked fine a week ago, I expect it to start or allow me to enter the editor.

Actual behavior
When I tried to enter the editor or start a scenario or preset version of the map, I got an "Error loading paths." error. When trying to start the main map the error "Unable to load game data!" appears.

Steps to reproduce the behavior
Try to start the map "War of Alagaesia - Perfected" or any scenario/preset of it.

Information:
Browser: Google Chrome

Browser version: Version 72.0.3626.119 (Official Build) (64-bit)

OS: Windows 7 Home Premium 64-bit

Any other relevant information: I'm the creator of the map and havn't changed anything since the last game. Perfectly playable and editable 8 days ago.

Attachments:
Game link: https://atwar-game.com/games/?link=4646314027
Error when trying to start the main version: https://media.discordapp.net/attachments/541104509557145624/554602503770210325/unknown.png
Link to the map preview (also broken): https://atwar-game.com/map/?id=22501

Confirmed.

Geschrieben von Zoyarox, 11.03.2019 at 06:34

Are you sure you haven't edited your map recently? On the server it was last modified on 2019-02-25....

EDIT: Btw I fixed your map, but the problem that caused it remains unknown. So this is still an open bug.

Technical info (for Clovis/Dave):

Problem is that somehow a map's versions folder on disk is advanced by 1, but version_paths in the DB is not. For example on this map, the versions folder changed from 3 to 4, but version_paths in DB stayed 3. I've noticed this issue with other maps also.... I don't know what's causing it, but it happens to other maps too, not just this one.

I pulled version 3 files from backup and did a comparison with version 4... path files were different, so it does seem the map was edited around the time the problem occurred. I have both versions saved offline for further debugging later...

The timestamps are interesting:

the versions/4 folder shows last modified 2019-02-25 01:58 (server time)
flat.zip last modified 2019-02-25 08:58 (but mappaths.xaml inside has time 07:58)
MapPaths1.xaml 2019-02-25 08:58
MapPaths2.xaml 2019-02-25 08:58
MapPaths3.xaml 2019-02-25 08:58

Strange that all timestamps end in :58, but different hours. Server timezone is UTC-6, so 01:58 + 6 = 07:58? But where did 08:58 come from? User's timezone? I don't know if the timestamp is even relevant, it just seems strange they aren't all the same.

Geschrieben von Dave, 11.03.2019 at 09:52

Geschrieben von Zoyarox, 11.03.2019 at 06:34

Are you sure you haven't edited your map recently? On the server it was last modified on 2019-02-25....

Thank you very much, Dave! <3
Yes I did edit it twice around that time. First, I slightly adjusted some borders (carefully to not get the Finalizing Error) and drew 2 no-go areas. Secondly, I changed some unit icons to the new ones.
But interestingly enough it all worked well in two games I had the day after and the following day! The new icons worked. Though I think the borders and no-go areas weren't applied in-game yet, while they were in the editor.

Geschrieben von Dave, 11.03.2019 at 10:01

Technical info (for Clovis/Dave):

Problem is that somehow a map's versions folder on disk is advanced by 1, but version_paths in the DB is not. For example on this map, the versions folder changed from 3 to 4, but version_paths in DB stayed 3. I've noticed this issue with other maps also.... I don't know what's causing it, but it happens to other maps too, not just this one.

I pulled version 3 files from backup and did a comparison with version 4... path files were different, so it does seem the map was edited around the time the problem occurred. I have both versions saved offline for further debugging later...

The timestamps are interesting:

the versions/4 folder shows last modified 2019-02-25 01:58 (server time)
flat.zip last modified 2019-02-25 08:58 (but mappaths.xaml inside has time 07:58)
MapPaths1.xaml 2019-02-25 08:58
MapPaths2.xaml 2019-02-25 08:58
MapPaths3.xaml 2019-02-25 08:58

Strange that all timestamps end in :58, but different hours. Server timezone is UTC-6, so 01:58 + 6 = 07:58? But where did 08:58 come from? User's timezone? I don't know if the timestamp is even relevant, it just seems strange they aren't all the same.

Interesting info!
Now I would've thought it'd be the changed borders or the added no-go areas that were applied later that caused the bug. But according to those timestamps everything should have been saved both for the game on the 26th and on the 27th.
Apparently that change caused the bug, but then I already shouldn't have been able to start any of those games back then...

The border change and the two no-go areas are still present in the editor version, but also still not in-game. (doesn't matter to me, but maybe that sheds some more light into the case)

I have no idea whats going on