DataMigrations.jl uses the dom_to_graph function to convert a category from a presentation to a graph so that it can be fed into the Limits functionality. At the same time it has to do some typecasting to handle the fact that its diagrams might containing at least three concrete types--SetFunctionCallable,SetFunctionId,FinDomFunctionInt--while also allowing for mutations during the process of populating partially defined diagram and then constructing bipartite graphs for the limit.
This PR modifies dom_to_graph to ensure that the output will always have dom a FinCatGraph, ob_map a vector of obtype, and hom_map a vector of homtype. There seems to be no reason as yet to allow codom(F) to be modified, although maybe there will be such a need someday.
Requesting a patch release. In some sense this is a breaking change since the old function wouldn't change a functor with domain other than a FinCatPresentation, but I don't have any reason to believe this function's ever been called outside of the DataMigrations.jl context. And the breaking could only occur if somebody was relying excessively on implementation details of the input functor such as the ob_map being a Dict anyway--this function doesn't change the functor's semantics.
DataMigrations.jluses thedom_to_graphfunction to convert a category from a presentation to a graph so that it can be fed into theLimitsfunctionality. At the same time it has to do some typecasting to handle the fact that its diagrams might containing at least three concrete types--SetFunctionCallable,SetFunctionId,FinDomFunctionInt--while also allowing for mutations during the process of populating partially defined diagram and then constructing bipartite graphs for the limit.This PR modifies
dom_to_graphto ensure that the output will always havedomaFinCatGraph,ob_mapa vector ofobtype, andhom_mapa vector ofhomtype. There seems to be no reason as yet to allowcodom(F)to be modified, although maybe there will be such a need someday.Requesting a patch release. In some sense this is a breaking change since the old function wouldn't change a functor with domain other than a
FinCatPresentation, but I don't have any reason to believe this function's ever been called outside of theDataMigrations.jlcontext. And the breaking could only occur if somebody was relying excessively on implementation details of the input functor such as theob_mapbeing aDictanyway--this function doesn't change the functor's semantics.