Closed escape0707 closed 3 years ago
I'm not sure why this is happening; the list here is showing 456.71 as the latest version.
Weird. I just looked the documentation of checkver
mechanism and it seems that the Nvidia download html page gives wrong order or wrong version at some point. Is there a way to dump the html when a downgrade of checkver
happens?
I don't think so, but I'm not well-versed enough in GitHub Actions to know for sure.
Okay, I'll try to look into it.
Oh, thank you. You're a legend. 😄
I looked up the debug log of some recent github actions. The action defined by https://github.com/Ash258/Scoop-GithubActions does say it sometimes finds 456.55 as the "latest version", but most of the time it do thing correctly.
I don't think I'll dive into Ash258's repo that soon. I'm write a python script and setup scheduling task to download and compare the html linked by checkver
:url
every hour, just before the github action started.
If this wrong behavior happens again and I caught the wrong html page, too. Then it's a Nvidia problem and we might have to find some other checkver
options. If the html I get just before the github action is correct but still the github action updates the repo with wrong info, then we can't avoid diving into Ash258's code.
Actually, I didn't catch any NVIDIA side HTML problem.
This bug isn't always happening at each time GitHub action Excavator
runs. Looking at the the log and the source code of the action, the problems happens when this script executes. It setups a lot of environment variables then invoke scoop
's auto-pr.ps1
. And the latter script invoke checkver.ps1
, then responses with a false "new version".
When I run checkver.ps1
on my machine, I always get a correct latest version number. Can't easily reproduct the problem on my side. I think setup Ash258
's script according to what he wrote and debug locally might be needed. To do the debugging, changing Ash258's script or even the original checkver.ps1
downloaded by Ash258's script is required, too. The problem don't always reproduce either. So the related knowledge scope of the bug is beyond that I expected. -_-
I have some questions for now.
First, is this repo using the correct version of Ash258's script? He suggested stable-win
branch but what's the meaning of the master branch?
Second, I see the Excavator
is setup to use Shovel
perhaps that needs to be checked for potential problems, too?
I don't think I could figure all these out now, have to prepare for recent exams....
Yes, I'm using the stable-win
branch and Shovel. The master
branch is no longer supported and runs on Linux rather than Windows. I don't know if Shovel is particularly reliable, but I needed it for my spotify
bucket, so I swapped over here to keep things consistent. I'll try disabling Shovel to see if it changes anything.
I also have exams coming up, so if that doesn't work, I'm not able to dedicate a lot of energy to this.
Thanks for looking into this!
The issue still persists without Shovel. 😑
Strangely, the issue seems to be resolved.
TinyNvidiaUpdateChecker hosts a website which gets the correct version.
Please review recent commits made by bots.