<div dir="ltr">But tell me how will you test your code that is depending on a class provided by external party? You cannot subclass so you cannot override the behaviour that you need stub, at the you will wrap entire library in your own classes that you can subclass but what for? Subclassing rarely breaks things and if it does, you should mark it final but it shouldn't be explicit.<div><br></div><div>Tomáš <div class="gmail_extra"><br><div class="gmail_quote">On Mon, Dec 21, 2015 at 6:31 PM, Javier Soto <span dir="ltr"><<a href="mailto:javier.api@gmail.com" target="_blank">javier.api@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I think it's just as important for methods. If you make a class final, all methods become final, so that's OK. But if you make a class subclassable, and then forget to mark some of it's methods final, then all methods would be overridable which is probably not what you'd want in must cases. <br><div><div><div class="gmail_quote"><div dir="ltr">On Mon, Dec 21, 2015 at 9:19 AM Tomáš Linhart <<a href="mailto:tomas@linhart.me" target="_blank">tomas@linhart.me</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div class="gmail_extra"><div>Hello,</div><div><br></div><div>I must say, I am not big fun of this proposal because currently in Swift only way how to mock classes is to subclass them. If this proposal becomes reality, it will make mocking of all third-party libraries impossible unless they mark their classes non-final and I am afraid authors will just use the default behaviour so at the end people will stop testing code that is using third-party libraries or they will have to fork the libraries or ask the authors. This can be fixed by having better testing support in Swift but I don't think, this will happen anytime soon.</div><div><br></div><div>I would rather see introduction of better reflection so mock frameworks can be reality. I would like to see also other building block of objected-oriented-programming such as abstract classes, protocols with generic type parameters and not just abstract types (associated types) that allows to design better APIs that don't depend so much on overriding regular classes.</div><span><font color="#888888"><div><br></div><div>Tomáš</div></font></span></div></div>
</blockquote></div></div></div><span><font color="#888888"><div dir="ltr">-- <br></div>Javier Soto
</font></span></blockquote></div><br></div></div></div>