https://bugzilla.redhat.com/show_bug.cgi?id=1047510 --- Comment #11 from Adam Williamson <awilliam@xxxxxxxxxx> --- Gregor: I'm still buried deep in autoloader theory, but I'm starting to think my approach is bad and yours is good. The convention seems to be to be relatively conservative with autoloading: only use it on classes that are explicitly registered with it, and have the autoloader implement some kind of prefix handling rather than using the include_path. I'll probably send an alternative PR which is closer to your approach, but tries to be efficient for the case where upstream's 3rdparty directory is present. I'm currently studying the Composer autoloader, which is quite an advanced one, for hints. I *do* think any reasonable autoloader implementation ought to try loading the PSR-0 (or PSR-4! look that one up :/) class path relative to the defined include path as a fallback if it can't find the class any other way, which is roughly the approach you took. At least, this morning I do. Maybe this afternoon I'll find another layer of this onion and change my mind :/ -- You are receiving this mail because: You are on the CC list for the bug. You are always notified about changes to this product and component _______________________________________________ package-review mailing list package-review@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/package-review