Thats a great set of observations Richard - I think I'd want it to work like you have described too - probably showing an inherited method in a light grey to indicate its inherited (but dark enough so you can compare it to the one on the other side.
I could potentially forsee an option where you could "show all inherited" methods on both sides to check how close objects are (in this case, probably up to the point where the inheritance trees unite - for many thats object). Another interesting (though experiment) option is to also show overriden methods and make space for things to keep lining up. Not totally sure on this - but thinking it might reveal accidental overrides or expose smells (again - just thinking aloud though). There's lots of interesting room to explore things in this tool, Tim p.s. Agree on the scroll behaviour and also the jump to next. On Thu, 25 Jul 2024, at 2:19 PM, Richard O'Keefe wrote: > OK, so I tried Class-diff in Pharo 11. I noticed two issues. > (1) I expected it to work like kdiff3, where scrolling scrolls both panes. > I'm so used to flying around differences using > Next/Previous/Last/First Difference buttons that I do very little > scrolling in kdiff3, but when I do scroll, it does its best to keep > the two panes aligned. > (2) Comparing Array and String, I eventually figured out that I was > being shown selectors *defined* in each class, but no selectors > *inherited* by a class. For example, String canUnderstand: #asArray > but inherits it, so it's not shown in the String pain, while Array > defines it. This *looks* like a "one class has it and the other lacks > it" case but it's *actually" a "both classes have it and it's > different in each", so a difference I'd really like to see. > I tracked this down to > self aligner > align: self sourceClass selectors sort > with: self targetClass selectors sort. > in updateAlignment, Changing that to > self aligner > align: self sourceClass allSelectors asArray sort > with: self targetClass allSelectors asArray sort > got me closer to what I wanted, but then accessing the source code of > inherited methods didn't work because inherited methods, by > definition, aren't in the class's own method dictionary. I could work > around that too, but it's 1:19 AM where I am. In fact that's not > quite what I want anyway. I don't want to be shown methods inherited > by *both* classes from Object (although methods inherited from Object > by one class and redefined or inherited from another class by the > other are ones I do want to see). > > On Wed, 24 Jul 2024 at 14:06, Richard O'Keefe <rao...@gmail.com> wrote: >> >> Thanks for the advice about Spec2. >> I was keen to try Class-diff out, so I fired up Pharo, followed the >> installation instructions with no problems, and then tried the >> example. >> The example got a run-time error, and I expect that it's because the >> Pharo release I fired up was Pharo 9. >> >> Which release(s) of Pharo is Class-diff intended to work with? >> >> On Tue, 23 Jul 2024 at 23:38, Hernán Morales Durand >> <hernan.mora...@gmail.com> wrote: >> > >> > >> > Thanks for your nice feedback Richard, I will sketch something and see how >> > it goes. >> > >> > About learning Spec2, for me, the best resource is to directly use the >> > image examples. In the main menu, you have "Help -> Spec 2 demo" and "Help >> > -> Spec 2 examples". And you also have the book in >> > https://github.com/SquareBracketAssociates >> > >> > Hernán >> > >> > >> > El mar, 23 jul 2024 a las 10:30, Richard O'Keefe (<rao...@gmail.com>) >> > escribió: >> >> >> >> That's a very nice tool. There are two tweaks I'd like to see. >> >> Two classes having methods with the same *name* isn't the same as two >> >> classes having the same *method*. >> >> When the two classes have a method with the same name, there are at >> >> least three possibilities: >> >> >> >> - The methods are the same method >> >> >> >> - The methods are different, but the name is defined in some common >> >> superclass, so they had better be related methods. >> >> >> >> - The methods are completely unrelated, like #next in ReadStream and >> >> LIFOQueue. >> >> >> >> The third case applies to every method. Perhaps the best way >> >> therefore is to show >> >> something in the header, like >> >> >> >> (score) diff of descendants of <nearest common superclass> >> >> <column> <column> >> >> >> >> The first and second cases are the ones I want to tell apart so that I >> >> don't spend any time looking >> >> for differences that aren't there. >> >> >> >> What's the best way to learn how to use Spec2? >> >> >> >> On Mon, 22 Jul 2024 at 18:58, Hernán Morales Durand >> >> <hernan.mora...@gmail.com> wrote: >> >> > >> >> > Dear Pharo community, >> >> > >> >> > I am happy to present a new tool designed specifically for comparing >> >> > classes in Pharo. It provides a two-sided list of methods, so you can >> >> > quickly understand the relationships between two classes. >> >> > >> >> > The project location is on GitHub: >> >> > >> >> > https://github.com/hernanmd/class-diff >> >> > >> >> > Feel free to send comments and contributions via PRs. >> >> > Have a great day, >> >> > >> >> > Hernán >> >> >
