On Thursday, May 2, 2013 13:12:55 Martin Briza wrote: > clear enough due to my English): this doesn't break anything nor introduce > any kind of regression, the functionality stays the same.
Well, code speaks louder than words in any language. I would suggest creating a branch to do this work in and start writing code. We can look at it as you progress and figure things like this out as we go. .. and I just read your follow up email saying you are doing exactly that. Awesome. :) > Yes, I totally agree with the abstract base class idea, though I still > think having a private subclass is a bit messy practice. Yet once again, It is less messy than having a publicly exposed factory with every user of the library having to deal with that implementation detail. "Messy" library internals are often necessary to make an elegant API. It also would mean we don't have to port any code that currently uses this API which is always a bonus: less work, fewer regressions. -- Aaron J. Seigo
signature.asc
Description: This is a digitally signed message part.
