{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifjw6fbylrlun577dpnhkrvwnmptvlbbjcxsroebvslfnye6lga5y",
    "uri": "at://did:plc:ivbknywyskln22er3nkssdhl/app.bsky.feed.post/3mjufpifjx2q2"
  },
  "path": "/t/matches-index-july-2025-reviving-the-left-pad-index/23272#post_17",
  "publishedAt": "2026-04-19T16:38:03.000Z",
  "site": "https://internals.rust-lang.org",
  "tags": [
    "(click for more details)"
  ],
  "textContent": "I wonder why the needs & wants are so heterogeneous.\n\nMy guess is that this is in no small part due to how we think of decimal numbers. Unlike for subsets of the integers, where we can clearly delineate the borders of the set at both ends (`::MAX` and `::MIN`), we can't do that for any subset of the rationals (or reals) since the concept itself means that they form a dense set.\n\n(Aside) **Dense** here in practice means that in any subset `S` of `P` (`P` here being e.g. the rationals or the reals), if we have two elements `x, y` in `S` such that `x < y`, the element `z = (x + y) / 2` will be such that `x < z < y`, and `z` will always belong to `P`. And if `S` is finite, this means that this operation will eventually take us out of `S`; much like how one can go outside of the floats just by finding space between two of them.\n\nWhat if floats weren't fixed width...? (click for more details)\n\nAnyway. Floats are kind of fucked. My guess is that half of this is because they're finite-width, which breaks our intuitions about decimals. The other half, we can probably fix within them, though. Well I hope so anyway x3",
  "title": "Matches-index (July 2025) -- reviving the left-pad index"
}