Re: [CDBI] Re: how can I make data returned from CDBI methods be considered untainted?
Michael G Schwern <[email protected]>
| Newsgroups | gmane.comp.lang.perl.modules.class-dbi |
|---|---|
| Message-ID | <[email protected]> |
Edward J. Sabol wrote: > Michael Schwern wrote: >> Argh! That's a private method (or at least protected)! > > I prefer the term "undocumented." :-) That's three times the reason not to use it! Whether its protected or private, you reached for a shotgun when bug spray will do. > Seriously, if I may be so presumptuous > as to debate the point with the original author of both Ima::DBI and > Class::DBI, I would say it's protected, not private. Based on some posts by > Tony on the old CDBI mailing list, my interpretation of the Class::DBI API, > which is certainly not unimpeachable, is that only methods that start with > double underscores are strictly forbidden (private). Methods with single > underscores are "mostly safe" (protected, sort of), though not absolutely > guaranteed to remain unchanged from version to version, of course. I'm not willing to comment on the naming conventions, but if its undocumented its undocumented and can change at any time. Without a documented API you override at your own risk. > _default_attributes() is inherited from Ima::DBI though, as I'm sure you > know, and then overriden in Class::DBI, which then calls the same method in > its superclass (Ima::DBI); it's very clearly not going anywhere. CDBI and Ima::DBI have a "special" relationship that's probably illegal in most US states. They're maintained by the same author and often released at the same time. Therefore, its not exactly the model for encapsulation. I myself am trying to eliminate such coupling from my own modules. I figure whatever guts I find myself poking at will be useful to someone else. I've mostly torn Test::More and Test::Builder's unseemly embrace apart.