Calculating any decomposition for integer matrices already fails to compile with a static assertion THIS_FUNCTION_IS_NOT_FOR_INTEGER_NUMERIC_TYPES. Shouldn't we do the same for triangular solvers? Would that break any (valid) code?
If the diagonal is filled with 1 (including UnitDiag mode), then this is valid, though I don't have use cases in mind.
Ok, that is true and might be a valid use case. And of course there can be other cases where all integer divisions happen to be exact. Maybe we should just document the current behavior (including that it can lead to uncaught divisions by zero).
-- GitLab Migration Automatic Message -- This bug has been migrated to gitlab.com's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.com/libeigen/eigen/issues/1639.