<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Adding Real-time Capabilities to a SML Compiler · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1601.03116/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/1601.03116/feed.xml" rel="self" type="application/rss+xml" /><item><title>Adding Real-time Capabilities to a SML Compiler — mentioned by twotwotwo</title><link>https://news.ycombinator.com/item?id=12059913</link><guid isPermaLink="false">urn:hackerpapers:mention:1601.03116:12059913</guid><pubDate>Sat, 09 Jul 2016 03:13:17 GMT</pubDate><description>&lt;p&gt;Via https://twitter.com/silentbicycle/status/751574435199672321, I found http://arxiv.org/abs/1601.03116, which talks about how they do the realtime GC stuff. An interesting bit for me is Section 4&amp;#x27;s way of handling fragmentation. They have a hard max size on allocations: large objects are linked lists and arrays are managed as trees (32-ary trees, so lookups are often not super deep). There&amp;#x27;s an obvious price in locality and pointer-chasing, but if consistent pauseless GC (and not throughput) is your goal, never having to compact to make space for a big allocation is neat.&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1601.03116/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.PL</category></item></channel></rss>
