It obviously also means that all the other tag values have to use 1 as the last bit.
Another reason why we wouldn’t want to use the top bit is that, as the parent comment suggested, the tagged pointer representation of a fixnum integer isn’t a pointer at all but is instead twice the number it represents. Generally speaking, we represent integers in twos-complement representation which uses that top bit to determine if the value is positive or negative.
Would love to see recent work demonstrating this isn’t true!
I read both articles and am looking forward to your next! I’d be interested in understanding more about the relationship of EMacs to GCC since you noted the authors were the same and the internals were written with some compiler awareness.
You made a great point. Since the original authors are the same, the fundamentals of the Emacs C core are indeed highly compiler-optimized. I hope I can come up with a way to fully understand and write about that history and relationship. (But to be honest, diving into that level of compiler history is a really hard topic to tackle!)
Thanks for the great inspiration and for taking the time to read!
I (briefly) searched for more info but this SO answer[0] implies they're quite similar. I'd be interested in understanding more. I feel like much of computer science has normalised to specific approaches that 'won' in the industry, where Lisp is an example of something interesting that largely didn't win -- with notable exceptions like HN itself! [1] :) And so hearing about uses of things like Lisp concepts in areas we might not know they're being used would be wonderful HN content.
[0] https://stackoverflow.com/questions/40799696/how-is-gcc-ir-d...
Symbols are just list of numbers. Variable is just a nameless place in memory, but often associated with a symbol.
Numbers in symbols are printed out as ASCII-characters when it seems appropriate, like after SETQ.
Or we could decide that number-list that ends with 0 and contains only range(0x21,0x7F) is printed out as symbol. Does not matter, it is just syntactic sugar.
And We do not need strings for much anything. We could of course decide that number-list with ord('"') is printed out as string. The reader could also follow this protocol.
I had all this figured out at one time. And I dont remember any major issues. B-)
I love Emacs' design. I think the cruft is minimal, and pretty much every line of code I studied had a good reason for being there.
And I also think there's a lot to learn from studying how Emacs is implemented. Few people will probably do this, but it was a nice experience for me. I learned a lot about gaps, text properties, how buffers are implemented, how the eval function works (it's surprisingly complicated between buffer-local variables and thread-local variables, but it's hard to think of a simpler alternative), and how intervals are implemented.