Endless loop in cookbook.(*CookbookVersion).resolveDependencies if cookbooks have circular dependencies about goiardi HOT 9 CLOSED

Yeah, the depsolver's always been a little weak. I'm surprised that this case (along with the zero byte file in the other issue) isn't covered in the chef-pedant tests, but here we are. I was planning on fixing the depsolver anyway, so I may as well finally get that out of the way too.

from goiardi.

At last, I have the rewritten depsolver in a state where I believe it's suitable for testing again. It's now sending back errors in the format chef-pedant expects, although some of the error messages need further refinement. More importantly here, it's more robust and shouldn't blow up like you had happen anymore. Of course, since the depsolver's a pretty touchy piece of the software, testing is appreciated.

from goiardi.

Awesome! I'll test it today.

from goiardi.

I get an error in chef-client:

https://gist.github.com/oker1/6bc2110838d4a1fdd966

I'll try to dig into a bit more.

from goiardi.

If you haven't, could you run that with debug on? I have a feeling I know where that's coming from, but if you could try running it with more debug output that'd help narrow it down.

from goiardi.

I've been getting back in the goiardi saddle recently, and just duplicated this bug. This should be figured out fairly soon.

from goiardi.

Well, that was a stupid error on my part. I pushed a fix for this up on the 0.9.1-dev branch. Urk.

from goiardi.

And even better, it looks like the circular dependencies no longer makes the memory usage explode. Huzzah! @oker1, if you see this, I believe this problem is finally fixed for you. Sorry for the delay; I got super busy with vacation and various non-goiardi things, and couldn't work on it for a little while.

from goiardi.

The fix for this is now in master, and seems to work well.

from goiardi.