I wanted to make sure we were in agreement around an offline discussion of #619. I think I'll abandon the changes in that PR and just start a new one, once we reach agreement in this issue.
The thinking, as I recall, was to add a few new error codes:
E_UNREACHABLE_CELL: when an algorithm was unable to reach a desired cell. these situations may relate to pentagon distortion or cells being too far apart. this situation might be overcome with a "safe" version of the function or with future algorithm improvements (check the function documentation)
E_PENTAGON_DOMAIN: when attempting to access the deleted subsequence of a pentagon
E_UNEXPECTED_CODE_PATH: for when we reach a situation which we expect to be impossible. users should let us know whenever they see this, as this indicates an implementation bug
We would also remove E_PENTAGON, and use either E_UNREACHABLE_CELL or E_PENTAGON_DOMAIN in its place.
We would not addE_TOO_FAR, as that would be covered by E_UNREACHABLE_CELL.
We would also not addE_RETRY_WITH_SAFE, as that information will can be given in the specific function documentation.
One alternative
We could potentially split E_UNREACHABLE_CELL into two different errors:
E_TOO_FAR
E_PENTAGON_DISTORTION
And E_PENTAGON_DOMAIN would remain. My thinking is that it is distinct from E_PENTAGON_DISTORTION. Folks agree?
I wanted to make sure we were in agreement around an offline discussion of #619. I think I'll abandon the changes in that PR and just start a new one, once we reach agreement in this issue.
The thinking, as I recall, was to add a few new error codes:
E_UNREACHABLE_CELL
: when an algorithm was unable to reach a desired cell. these situations may relate to pentagon distortion or cells being too far apart. this situation might be overcome with a "safe" version of the function or with future algorithm improvements (check the function documentation)E_PENTAGON_DOMAIN
: when attempting to access the deleted subsequence of a pentagonE_UNEXPECTED_CODE_PATH
: for when we reach a situation which we expect to be impossible. users should let us know whenever they see this, as this indicates an implementation bugWe would also remove
E_PENTAGON
, and use eitherE_UNREACHABLE_CELL
orE_PENTAGON_DOMAIN
in its place.We would not add
E_TOO_FAR
, as that would be covered byE_UNREACHABLE_CELL
.We would also not add
E_RETRY_WITH_SAFE
, as that information will can be given in the specific function documentation.One alternative
We could potentially split
E_UNREACHABLE_CELL
into two different errors:E_TOO_FAR
E_PENTAGON_DISTORTION
And
E_PENTAGON_DOMAIN
would remain. My thinking is that it is distinct fromE_PENTAGON_DISTORTION
. Folks agree?Or is all this more detail than the user needs?