|
From: | Paul Eggert |
Subject: | Re: [Emacs-diffs] master 470082d: List lengths are always fixnums now |
Date: | Fri, 11 Jan 2019 11:10:30 -0800 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.2.1 |
Stefan Monnier wrote:
What would such a platform look like? I guess a CrayI-style platform where `sizeof (void*) == 1` and hence `sizeof (struct Lisp_Cons)` can be 2? Are there less hypothetical cases we could mention in the comment?
Now that I think about it, I was more worried about architectures where pointers (and therefore EMACS_INT) are wider than ptrdiff_t. Something like the IBM i:
https://www.ibm.com/support/knowledgecenter/en/ssw_ibm_i_73/rzarh/iptrtypes.htmThese systems are still in production, although highly proprietary. I don't know much about them.
Perhaps I should have written the summary as "List lengths are always ptrdiff_t now" as that was more the focus of that patch.
Getting back to your question, these days a word-addressible machine would probably be a DSP, something like a TI TMS320C3x. Admittedly an unlikely target for Emacs, and to some extent this part of the patch is more a comment on one of the (many) problems we'd have if we really wanted to port to a word-addressable machine.
[Prev in Thread] | Current Thread | [Next in Thread] |