Threadwell

Recent Notes

NOTimothyLottes
Aug 25, 2026
cmuratori
Aug 25, 2026
Threads
Aug 25, 2026
- archive

rfleury

72 items with this tag.

Aug 25, 2026
rfleury
Aug 07, 2026
A good way to evaluate one axis of quality of an abstraction is to inline every function call, and see how much duplicate work you can either deduplicate or turn into a loop (rather than several independent function calls).
Jun 02, 2026
The first way people usually learn code compression is by pulling common things out into helper functions, but in my view the complexity of adding a new entry point is often overlooked, and things quickly turn into a mess.
Apr 30, 2025
To the degree this is true, it’s only in a technical sense.
Dec 18, 2024
One of the most powerful lessons I learned as a programmer was to use combinatoric spaces to my advantage, rather than allowing them to explode into—for example—handwritten source code.
Aug 09, 2026
@rfleury makes me wonder why we don't have a tool (do we?) to visualize inlining like that (without actually doing it).
Aug 08, 2026
@rfleury John Carmack wrote about this: http://number-none.com/blow/john_carmack_on_inlined_code.html
Aug 08, 2026
@rfleury Some people will say that a good abstraction lets you understand the loops underneath, and some will say that a good abstraction means you don’t have to understand the loops underneath.
Aug 08, 2026
@rfleury The fundamental property of abstraction is to avoid repetition.
Aug 07, 2026
@rfleury If you got a concrete example of this, it might be worth making a video out of that, just saying! 😋
Aug 07, 2026
@rfleury Overall I’ve noticed that inlining is just amazing at surfacing issues with what the actual „abstraction” is - kind of that if I carve the function too early I’ll obfuscate the fact that maybe the actual function (behaviour) bends to that function rather than the other way around
Aug 07, 2026
@rfleury Assuming -O3, if you compare the assembly output before and after the exercise, do you see any meaningful difference?
Aug 07, 2026
@rfleury Do you have a favorite example from raddbg to look at?
Aug 07, 2026
@rfleury This seems like such good advice and hard to visualize without having practiced a couple of times.
Aug 07, 2026
@rfleury This is the first thing I do in every big refactor
Aug 07, 2026
@rfleury btw this sounds like a nice feature for a debugger
Aug 07, 2026
@rfleury couldn't resist
Aug 07, 2026
@rfleury I did this process to linearise the mutually recursive eval loop of a language - also found a very nice opportunistic TCO due to it.
Jun 03, 2026
@rfleury will love to read a long article on this
Jun 03, 2026
@rfleury Can you share an open source project that follows this strategy? As someone deeply entrenched in the web and java world, I am finding it hard to wrap my head around.
Jun 03, 2026
@rfleury rad debugger is easy to read for such a complex project, I'm experimenting with the same long proc and noticed that when I add something I'm not constrained to early abstractions, I can connect dots much easier.
Jun 03, 2026
@rfleury I like a big ass on my functions.
Jun 03, 2026
@rfleury Interesting! I was just reviewing code and thinking of solution when I read this tweet and realised "table + removing enum" solved it better than expecting the helper function to be used correctly.
Jun 03, 2026
@rfleury Do you have a code example of how you would compress / simplify something?
Jun 02, 2026
@rfleury I like using lambdas or local functions to extract some common logic and put it behind a good name without cluttering things up outside
Jun 02, 2026
@rfleury yes! A similar problem comes when you follow retarded bob martin-esque rules like not exceeding 6 lines per function.
Jun 02, 2026
@rfleury I completely agree, adding a helper function rarely reduces complexity and often leads to a lot of argument bloat.
Jun 02, 2026
@rfleury oh I hate helper functions, especially the ones that get used exactly once and serve no purpose but hiding code
Jun 02, 2026
@rfleury I've come to refer to declarative as "just fucking say it" coding.
Jun 02, 2026
@rfleury I've been compressing in terms of "section functions" as opposed to helper functions.
Jun 02, 2026
@rfleury I noticed focusing on the data you work and how you work on it helps a lot
Jun 02, 2026
@rfleury I'll often inline every 'helper function' I find while coding.
Jun 02, 2026
@rfleury Entry point you mean indirection ?
Jun 02, 2026
@rfleury fewer entry points = fewer surprises
Jun 02, 2026
@rfleury I'm not exactly sure what do you mean with this statement.
Jun 02, 2026
@rfleury I like giant matching functions with lots of helper functions as closures with limited scope.
Jun 02, 2026
@rfleury Only way to compress is dsl vv everything else is just the dsl with some extra boilerplate
Jun 02, 2026
@rfleury 100% agree, DRY can be extremely overdone.
Jun 02, 2026
@rfleury Do you have an example or a before and after by chance?
Jun 02, 2026
@rfleury Very well said.
Jun 02, 2026
@rfleury My mental model for this: Are you putting the mess in a closet, or are you actually organizing the it?
Jun 02, 2026
@rfleury A function explicitly limits the variables that its code depends on.
Jun 02, 2026
@rfleury I like this as it maps the problem with input on one side and output on the other.
Jun 02, 2026
@rfleury Does this imply if you took an external dependency, like Linux or matrix library, and started folding that code into your own, instead of calling into it, your code would improve?
Jun 02, 2026
@rfleury Can't tell just how many times I've pulled out something into it's own function only to use it once.
Jun 02, 2026
@rfleury I got a mess (kind of working fine but dirty) with c# extensions.
Jun 02, 2026
@rfleury db 설계에 앞서 모든데이터를 파일 입출력 으로 해서 최소 부하를 해보는것도 도움이됨
Jun 02, 2026
@rfleury Never thought of helper f-ns as code compression, you still have all the same stuff elsewhere.
Jun 02, 2026
@rfleury What about polymorphism and inheritance?
May 01, 2025
@rfleury individual malloc/free is a common C practice and is the fair comparison; group allocations are not the norm in most programs especially in the hand-off ownership situations where unique_ptr would apply.
May 01, 2025
@rfleury Look at any large codebase and you'll see a bunch of new/delete calls.
May 01, 2025
@rfleury Do you have any example of this?
May 01, 2025
@rfleury Well said.
Apr 30, 2025
@rfleury Saw a Casey Muratori video about people who think in single unit lifetimes vs group allocations etc yesterday.
Apr 30, 2025
@rfleury There is exactly zero pain in debugging unique_ptr variables these days.
Apr 30, 2025
@rfleury I miss the point you are trying to make
Apr 30, 2025
@rfleury This isn't such a strong argument considering there's a "leak" with the regular call to `new`.
Apr 30, 2025
@rfleury Too much words, but unique_ptr is*really* zero cost.
Apr 30, 2025
@rfleury Generated assembly or other supposedly better ‘hand-written’ options for memory management are IMO just distractions when talking about unique_ptr.
Apr 30, 2025
@rfleury I bet std comitee wrote thousands of lines for this
Apr 30, 2025
@rfleury I'd use C more if attribute(__cleanup__) was standard and a little less cumbersome.
Apr 30, 2025
My experience has been different.
Apr 30, 2025
@rfleury Also people will say this garbage and then show you the assembly in -O2.
Jan 24, 2025
@rfleury thank you for this thread- read it a while ago, but now working on a game and something *clicked* when thinking about modeling data in this way (understanding the power of the 2^n benefit)
Dec 20, 2024
@rfleury interesting projection of combinatorics system into coding system.
Dec 19, 2024
@rfleury I completely agree.
Dec 18, 2024
@rfleury example?
Dec 18, 2024
@rfleury @threadreaderapp unroll
Dec 18, 2024
@rfleury At what point do you think you should convert a switch statement to "behavior flags"? Suppose there are N entity types and K behaviors.
Dec 18, 2024
@rfleury Maybe you already wrote this somewhere, but I'd be interested in seeing a practical example of your approach, even a basic one.
Dec 18, 2024
@rfleury any concrete example?
Dec 18, 2024
@rfleury This reminds me of games and simulations, where entities have a set of attributes.

Created with Quartz v5.0.0 © 2026

GitHub