Tuesday, March 30, 2021

Should have learned by now...

 ...to not say that things are going well. 

So I've been having some weird issues lately, problems that I've sometimes been unable to reproduce and they are seemingly random in behavior. I've mistakenly assumed these have been part of the conversion to using "several threads" described earlier.

But this time I noticed that some weird graphical bugs creeped in and since that code "cannot" do anything wrong (it basically sets a number on or off in a fixed array), I figured it must be a disk/flash issue since the letters are hardcoded . 

1) The M is consistently erroneous despite working fine the upload before etc.  


Then I realized, I hadn't enabled "Verify flash after upload" .. and, yeah. This. 

2) Verification always fails at the same adress on the installed Chipkit. 

Note that the flash is completed, so the machine boots and is playable. But it appears that whatever ends up in that space (this time it was the letter 'M') gets corrupted and could cause issues.

I've been uploading to the Chipkit a lot over the years, so it's not unlikely that the flash has finally failed. I also tested to do an upload and verify against a spare Chipkit MAX32 that I have, and that verified just fine.

I might try to keep uploading and trouble shoot on the broken (?) Chipkit until it completely breaks rather than swapping it now, since I'm working on isolated code segments at the moment.  

Update: 

The verification sometimes do succeed.  Obviously I won't leave things to chance later on, but for development of game scenes and thing that are "fixed" in scope - I can manage. 



No comments:

Post a Comment