I looked through the Python solutions again. The only incorrect solution I found is this one, but it only fails (i.e. incorrectly returns true) for a few inputs.
When tests are updated, only some amount of top solutions gets revalidated, and I can't remember if it's 50 or 500. It's never revalidating all solutions, because it's too expensive. Additionally the process is flaky, and when revalidation of one solution fails (what happens more often than we'd like it to), the revalidation process stops and remaining solutions are not checked.
Mods can revalidate specific solutions on demand.
Bottom line is, When there is more than 50 (or 500, can't remember) solutions, not all of them will be ever rechecked.
Can't tell for sure, it says that the task is scheduled. I verified that solution manually and now it's invalid. As far as I know, the only button there is, is for manual verification. The other thing we can do is to add and approve a new fork, but I'd only do that after a couple of days if the current one doesn't trigger.
My broken Python solution still hasn't been invalidated. How long does it usually take between approval of stricter tests and invalidation of incorrect solutions? Can someone push a button to (re)start the invalidation process?
Added it to the "too short" tests.
We could simply add
['s','s','s','s','s','w','w','n']tofailwalkand update the indices infixed_tests()...A couple of manual verifications didn't render that solution invalid, so you'll have to write a specific test to do that.
I looked through the Python solutions again. The only incorrect solution I found is this one, but it only fails (i.e. incorrectly returns true) for a few inputs.
Thanks for the explanations! By now, my broken solution has been invalidated, and several others as well, so I guess it's OK.
When tests are updated, only some amount of top solutions gets revalidated, and I can't remember if it's 50 or 500. It's never revalidating all solutions, because it's too expensive. Additionally the process is flaky, and when revalidation of one solution fails (what happens more often than we'd like it to), the revalidation process stops and remaining solutions are not checked.
Mods can revalidate specific solutions on demand.
Bottom line is, When there is more than 50 (or 500, can't remember) solutions, not all of them will be ever rechecked.
Can't tell for sure, it says that the task is scheduled. I verified that solution manually and now it's invalid. As far as I know, the only button there is, is for manual verification. The other thing we can do is to add and approve a new fork, but I'd only do that after a couple of days if the current one doesn't trigger.
My broken Python solution still hasn't been invalidated. How long does it usually take between approval of stricter tests and invalidation of incorrect solutions? Can someone push a button to (re)start the invalidation process?
I didn't look at the new Python tests, but I can try and translate them to F#. I guess that F# tests could use migration from Fuchu to NUnit too.
Thanks! There are always 6 tests among the full tests that fail for this code.
Approved the Python fork. Thanks. I think that resolves this issue, at least that code seems to fail even after 10 retries.
My F# experience is very limited, but if I have time in the next couple of days, I'll take a look.
I added Python tests for all 286 combinations of counts of n,s,e,w. This should invalidate most broken solutions, for example the one above.
Please have a look and approve the translation. Thanks!
I don't know if you do F#, but check the issue below if you do.
I'll add tests.
Loading more items...