Choosing a language based on its syntax? | Hacker News Reader

Choosing a language based on its syntax?

gingerbill.org

40 pointstodsacerdoti109 comments

Semantics are where the rubber meets the road, certainly; but syntax determines how readable the code is for someone meeting it the first time.

Contrast an Algol-descendant like C, Pascal, Java, or even Python with a pure functional language like Haskell. In the former, control structure names are reserved words and control structures have a distinct syntax. In the latter, if you see `foo` in the body of a function definition you have no idea if it's a simple computation or some sophisticated and complex control structure just from what it looks like. The former provides more clues, which makes it easier to decipher at a glance. (Not knocking Haskell, here; it's an interesting language. But it's absolutely more challenging to read.)

To put it another way, syntax is the notation you use to think. Consider standard math notation. I could define my own idiosyncratic notation for standard algebra and calculus, and there might even be a worthwhile reason for me to do that. But newcomers are going to find it much harder to engage with my work.

I absolutely agree about Haskell (and also OCaml). They both suffer from "word soup" due to their designers incorrectly thinking that removing "unnecessary" punctuation is a good idea, and Haskell especially suffers from "ooo this function could be an operator too!".

> Contrast an Algol-descendant like C, Pascal, Java, or even Python with a pure functional language like Haskell. In the former, control structure names are reserved words and control structures have a distinct syntax. In the latter, if you see `foo` in the body of a function definition you have no idea if it's a simple computation or some sophisticated and complex control structure just from what it looks like. The former provides more clues, which makes it easier to decipher at a glance. (Not knocking Haskell, here; it's an interesting language. But it's absolutely more challenging to read.)

For what it's worth, Python has been moving away from this, taking advantage of a new parser that can implement "soft keywords" like 3.10's "match" statement (which I'm pretty sure was the first application).

Believe it or not, the motivation for this is to avoid reverse compatibility breaks. Infamously, making `async` a keyword broke TensorFlow, which was using it as an identifier name in some places (https://stackoverflow.com/questions/51337939).

In my own language design, there's a metaprogramming facility that lets you define new keywords and associated control structures, but all keywords are chosen from a specific reserved "namespace" to avoid conflicts with identifiers.

I don't have any real problem with words that are reserved absolutely and words that are reserved just in particular places. My point was more that in Algol-derived languages control structures look like control structures. And even in languages that implement `map()` and other higher-order functions, you can tell that it's a method/function call and that these things are being passed to it and those other things are not without going and looking up what `map()` does.

> In the latter, if you see `foo` in the body of a function definition you have no idea if it's a simple computation or some sophisticated and complex control structure just from what it looks like.

All control structures are reserved as keywords in Haskell and they're not extensible from within the language. In C I can't tell that an if(condition) isn't a function call or a macro without searching for additional syntactic cues, or readily knowing that an if is never a function. I generally operate on syntax highlighting, followed by knowing that an if is always a control structure, and never scan around for the following statement terminator or block to disambiguate the two.

I've found in general programmers greatly overestimate the unreadability they experience with the ISWIM family to be an objective property of the grammar. It's really just a matter of unfamiliarity. Firstly, I say this as a programmer who did not get started in the ML family and initially struggled with the languages. The truth of the matter is that they simply engage a different kind of mental posture and have different structural lines you're perceiving, this is generally true of all language families.

Pertinant to that last point and secondly, the sense of "well this is clearly less readable" isn't unique when going from the Algol family to the ISWIM family. The same thing happens in reverse, or across pretty much any language family boundary. For example: Prolog/Horn clauses are one of the least ambiguous syntax families (less so than even S-expressions IMO), and yet we find Elixir is greatly more popular than Erlang, and the most commonly cited preference reason has to deal with the syntax. Many will say that Erlang is unintuitive, confusing, strange, opaque, etc. and that it's hard to read and comprehend. It's just the same unfamiliarity at play. I've never programmed Ruby, I find Elixir to be borderline incomprehensible while Erlang is in the top 3 most readable and writable languages for me because I've spent a lot of time with horn clauses.

I think there's a general belief programmers have where once you learn how to program, you are doing so in a universal sense. Once you've mastered one language, the mental structures you've built up are the platonic forms of programming and computer science. But this is not actually the case. More problematically, it's propped up and reinforced when a programmer jumps between two very similar languages (semantically and/or syntactically) and while they do encounter some friction (learning to deal without garbage collection, list comprehensions, etc), it's actually nothing that fundamentally requires building up an entirely different intuitive model. This exists on a continuum in both semantics and syntax. My Erlang example indicates this, because semantically the language is nothing like Prolog, its differentiation from Elixir is purely syntactic.

There is no real universal intuition you can build up for programming. There is no point at which you've mastered some degree of fundamentals that you would ever be able to cross language family boundaries trivially. I've built up intuition for more formal language families than is possibly reasonable, and yet every time I encounter a new one I still have to pour a new foundation for myself. The only "skill" I've gotten from doing this ad nauseum is knowing at the outset that mastery of J does not mean I'd be able to get comfortable reading complex Forth code.

> its differentiation from Elixir is purely syntactic.

Well, there's also standard library, Erlang one is very messy while Elixir one is very consistent (and pipe operator - `|>` - enforces order of arguments even in low-quality 3rd party code as well, making whole language more pleasant to work with. Same goes for utf8-binary string everywhere and other idiomatic conventions

> There is no real universal intuition you can build up for programming. There is no point at which you've mastered some degree of fundamentals that you would ever be able to cross language family boundaries trivially.

I don't really agree with you on this, even though I agree with everything else here. Then again, I am an outlier where I've used ~40 programming languages in my career. There are a couple of language families (array languages like APL, exotics like BF) where I cannot read it because I've had no real opportunity to learn them, and there's a significant difference in being able to read a language and use a language (I can read, but not really use Haskell -- although I have shipped a couple of patches to small libraries).

I despair at the number of developers in the profession who understand only one or two programming languages…and badly at that.

(It's worth noting that I wholly disagree with the original post. 24 years ago I chose Ruby over Python because of syntax. Ruby appealed to me, Python didn't — purely on syntax. I never pretended that Python was less capable, only that its syntactic choices drove me away from choosing it as a primary language. I'm comfortable programming in Python now, but still prefer using most other languages to Python … although these days that has more to do with package management.)

> All control structures are reserved as keywords in Haskell and they're not extensible from within the language. In C I can't tell that an if(condition) isn't a function call or a macro without searching for additional syntactic cues, or readily knowing that an if is never a function. I generally operate on syntax highlighting, followed by knowing that an if is always a control structure, and never scan around for the following statement terminator or block to disambiguate the two.

Any Haskell function can serve as a control structure in the Algol sense, not so? As for `if(test)` that could indeed be a macro if the programmer’s a durned fool; but absent macros I don’t believe it can be a function call.

Mind you, I take your point about language familiarity; obviously people manage it.

Syntax is what keeps me away from Rust. I have tried many times to get into it over the years but I just don't want to look at the syntax. Even after learning all about it, I just can't get over it. I'm glad other people do fine with it but it's just not for me.

For this reason (coming from C++) I wished Swift were more popular because that syntax is much more familiar/friendly to me, while also having better memory safety and quality of life improvements that I like.

Do you have some examples of what you couldn't get along with? I know this is a lot to ask, but to me while I do write Rust and I don't write C++ or Swift in volume (only small examples) the syntax just doesn't feel that different really.

If you do like Swift you might want to just bite the bullet and embrace the Apple ecosystem. That would be my recommendation I think.

This resonate so much to my relationship with Rust. Also with Go. I'm having hard time learning Rust's advacend concepts because of its syntax.

Strangely enough I find Lisp's parentheses much more attractive.

Swift's syntax may look nice, but as soon as you run into "The compiler is unable to type-check this expression in reasonable time; try breaking up the expression into distinct sub-expressions" you'll forget all of that. Hint: they are related.

It didn't have to be like that though... It hasn't always been this bad. Personally I think they've been ruining Swift since version 4.

It's particularly terrible in SwiftUI context nowadays but you can also make it chuck on something as simple as a .map(...)

Wow, this is one of the most surprising comments I've ever read on HN!

Personally, I bucket C++ and Rust and Swift under "basically the same syntax." When I think about major syntax differences, I'm thinking about things like Python's significant indentation, Ruby's `do` and `end` instead of curly braces, Haskell's whitespace-based function calls, Lisp's paren placement, APL's symbols, etc.

Before today I would have assumed that anyone who was fine with C++ or Rust or Swift syntax would be fine with the other two, but TIL this point exists in the preference space!

There are a lot of important points of difference. Whether functions are introduced with an explicit keyword; how return types are marked; whether semicolons can be omitted at end of line; how types are named (and the overall system for describing algebraic types). Not to mention semantics around type inference (and whether it must be explicitly invoked with a `var` or `auto` etc.) and, again, algebraic types (what means of combination are possible?). And specifically with Rust you have the syntax required to make the borrow checker work. Oh, and then there are the implicit returns. Rust certainly didn't invent that (I seem to recall BASIC variants where you could assign to the current function name and then that value would be returned implicitly if control flow reached the end), but it reflects a major difference in philosophy.

... Which is really all to say: different people are focused on different details, at different levels.

> Whether functions are introduced with an explicit keyword; how return types are marked; whether semicolons can be omitted at end of line; how types are named (and the overall system for describing algebraic types).

Yes, these are all examples of things I always thought were generally considered small enough differences that nobody who was okay with how C++ or Rust or Swift did them would find the way one of the others did it a deal-breaker.

> ...Which is really all to say: different people are focused on different details, at different levels.

For sure!

> examples of things I always thought were generally considered small enough differences...

The thing is that they add up. Writing C or C++ is unpleasant enough for me that I've seriously thought about learning Rust just to have an alternative in that niche that actually has traction this time. (I wasn't under the impression that Swift — or Go, similarly — is intended to be quite as low-level. But maybe they are?)

Definitely second this sentiment. Rust just... Looks wrong. And for that reason alone I've never tried to get into it.

I understand exactly how shallow that makes me sound, and I'm not about to try and defend myself.

No need to defend yourself, I share this sentiment as well. If I'm going to spend time writing and reading a lot of code in a new learning language, I want my previous knowledge to be somewhat reusable.

For this reason I was able to get into Odin as opposed to Zig because of some similarities with Swift Syntax as well how easy it is to parse.

The less I need to rewire my brain to use xyz language, the greater the chance of me getting into it.

If my life depended on it, I could get over such a shallow reason to dismiss a language but fortunately it doesn't and that's why I write Swift rather than Rust.

When I was a kid learning BASIC, a lot of beginner examples in books used the (purely decorative) keyword LET for every assignment. Consequently I associate it with "coding like a baby who understands nothing" and still hate to write let to this day.

The syntax of a language is the poetry form, it defines things like meter, scansion, rhyming scheme. Of course people are going to have strong aesthetic opinions on it, just as there are centuries of arguments in poetry over what form is best. You can make great programs in any language, just like you make beautiful poetry in almost every form. (Leaving an almost there for people that dislike Limericks, I suppose.) Language choice is one of the (sometimes too few) creative choices we can make in any project.

> Another option is to do something like automatic semicolon insertion (ASI) based on a set of rules. Unfortunately, a lot of people’s first experience with this kind of approach is JavaScript and its really poor implementation of it, which means people usually just write semicolons regardless to remove the possible mistakes.

Though the joke is that the largest ASI-related mistakes in JavaScript aren't solved by adding more semicolons, it's the places that the language adds semicolons you didn't expect that trip you up the worst. The single biggest mistake is adding a newline after the `return` keyword and before the return value accidentally making a `return undefined` rather than the return value.

In general JS is actually a lot closer to the Lua example than a lot of people want to believe. There's really only one ASI-related rule that needs to be remembered when dropping semicolons in JS (and it is a lot like that Lua rule of thumb), the Winky Frown rule: if a line starts with a frown it must wink. ;( ;[ ;`

(It has a silly name because it keeps it easy to remember.)

Language syntax is like the weather. When it's good (or when you're acclimated to it, I guess) you don't notice it. When the weather is perfect you don't even feel like the atmosphere even exists. When a language is so ingrained in your mental models, you don't even notice syntax, you just see semantics.

True to any design in general.

In the follow up article, I wrote something like this in the conclusion:

> Designing a language is more than just getting the syntax right, and denotation semantics are a lot more important in most ways, but that does not mean syntax does not matter. When you get syntax right, people won’t even know you’ve done anything at all. But when you’ve got it wrong (and have actually tried using the language), they will complain instantly.

Follow-up Article: https://www.gingerbill.org/article/2026/02/21/does-syntax-ma...

I am an S-exp enjoyer, and more for practical reasons than aesthetic ones—I really like the editor tooling that's possible with S-expressions. So I will absolutely choose a Lisp or a lisp if given the option, even at some level of inconvenience when it comes to the maturity of the language itself. I will always write Hy[0] rather than Python, for example.

[0] https://hylang.org

(I am aware of Combobulate[1] for Emacs folks, of which I'm sadly not one.)

[1] https://GitHub.com/mickeynp/combobulate

The most machine-friendly syntax - and the least appropriate for our LLM overlords which get confused by parenthesis because they don’t see the structure.

Have you tried? LLMs are really good at elisp, which is strange because elisp code is almost always GPL.

There's an option beyond lisp. Forth has even less syntax.

In my experience, they find counting parenthesis as difficult as counting the r's in strawberry if it was spelled srtrawrrrbrrerryrrrr.

I have really just one wish when it comes to syntax: no syntactically significant whitespace. Space, newline, tab, etc. should ALL map to the same exact token. In practice this also means semicolons or something like them are needed as well, to separate expressions/statements. I dislike langs that try to insert semicolons for you, but at least it's better than the alternative.

the way python treats whitespace is a huge design mistake that has probably wasted like a century (if not more) worth of time across all users, on something really trivial.

That's one of the things I like about C, the independence in how one can write code. I was able to develop my own style thanks to that, visualising the structure of the code to distinguish the different parts of statements and make it more clear (at least to myself).

(edited several times to try to correct changes in formatting for an example here, but it's just screwed up :-/ )

https://www.perlmonks.org/?node_id=45213

I agree, that I do not like automatic semicolon insertion (in my opinion it is one of the worst features of JavaScript, and possibly the really worst one), and I think it is a good idea that you should use semicolons or whatever to separate expressions and statements (except for a programming language where it is already unambiguous (e.g. because you are required to have brackets around them instead), in which case it is unnecessary).

However, spaces, line breaks, tabs, page breaks, etc are not normally tokens (and should not be tokens), but will separate tokens.

However, that is not the only issue with the syntax, although it is a significant one.

Yeah I misused the term "token". What I really meant is that they should just be token separator and nothing more (i.e. once we encounter whitespace it means the current token has ended)

This is the old "syntax does not matter" claim. Syntax is not the most important thing in the world when it comes to programming languages, but it does matter too. I was using perl, then PHP then ruby. There is no comparison here; ruby beats the other two languages hands down. I get to be able to do more, with less syntax and it is easier to read too (provided you write good code; you can write horrible code in any language of course).

Most of the languages that are created anew, end up being a clone of C or C++. Go is one of the few exceptions here; Rust is not an exception. It is basically C++ really, from the syntax - or even worse.

Sadly it is not possible to try to convince people who claim that syntax does not matter, that it does matter. They just keep on repeating that syntax is irrelevant. I don't think syntax it is irrelevant at all. It has to do with efficiency of expression. Clear thoughts. Clear design. It is all inter-connected.

> Rust is not an exception. It is basically C++ really, from the syntax - or even worse.

That's a superficial impression. Yes the resemblance is on purpose, but it doesn't go very deep which suggests you haven't stared at this for more than a few seconds.

I literally had to write an extra article because of the number of people who thought I was saying syntax doesn't matter. It matters a hell of a lot:

https://www.gingerbill.org/article/2026/02/21/does-syntax-ma...

I was trying to be as clear as I possibly could within that article to state I was only focusing on those two specific topics, rather syntax as a whole. How people judge a language based purely on its declaration syntax and nothing else. Where the declaration syntax makes the least difference compared to the rest of the syntax in the language. And still people wrongly interpreted the article.

An article about diversity of language syntax that somehow only deals with C-adjacent curly-brace languages (and,tbf, Odin).

This is a blinkered viewpoint. If you want to talk about syntax, at least mention the Haskell family (Elm, Idris, F*, etc), Smalltalk, and the king of syntax (less) languages, LISP (and Scheme), which teach us that syntax is a data structure.

The article does no such thing and just focused on declaration syntaxes and semicolons. I can class literally any language into the three categories for declaring variables.

* Haskells are name-focused languages. * Smalltalk is a name-focused language. * LISPs are qualifier-focused languages.

I think you might have a blinkered viewpoint in how you have interpreted the article.

> * Haskells are name-focused languages. * Smalltalk is a name-focused language. * LISPs are qualifier-focused languages.

First, I reject the idea that all languages can be classified according to this metric. The fact that you think that they can reveals how little experience you have with different languages.

Second, your proposed classification of these languages, even by the naive standards of the article, is wrong. You've proved my point.

First, tell me languages which have named value declarations which don't fit into that category. I'd love for my hypothesis to be proven wrong.

Second, how are they wrong? You haven't proved any point as you were extremely vague.

You're missing the point. Your categories are irrelevant. It's like if I announced proudly "All cars can be divided three categories: those with rear lights angled up, those with rear lights angled down, and those with rear lights straight." True but useless.

But if you're going to categorize, your categorization is wrong. Just because a language puts a qualifier first in no way means that it is qualifier focused.

The categorizes are useful for understand how people perceive a language. It's a human psychology category, and just because you think it is "irrelevant" misunderstands why I made the categorical distinctions in the first place.

People actually talk this way about languages, and they already have this conception.

And I've just realized what your psychological hang-up is now too, the word "focused". I originally didn't use this term when I wrote the article in 2018, but a friend suggested the term instead because it was probably a little more "neutral". I honestly cannot remember what I used before now since I have just absorbed this arbitrary term "focused", even if the distinctions are useful.

You might not think this way about languages. I know I don't when I program. But I know this is how others things, and that's the entire aspect of design. Design is just so much about understanding humans. How they function, mentally and physically. Perception, psychology, sociology, physiology, ergonomics, needs, desires, etc. It's all about being able to put yourself in other people's shoes, more than making _the thing_.

And that's what I've been trying to understand and that categorization has been helpful for understand other people. I don't care if you think the categorizes are "irrelevant" but you are not "everyone".

I think you oversimplify things a bit.

Are those names erased during compilation? It has a massive impact.

If you have indirect calls, how are those resolved? That matters a lot.

What is even the language, after the code is compiled/interpreted. Does it disappear like in many languages? Do you have some parts available, but not all (like in PHP)? Or do you have full runtime at hand and you can mold it like in Smalltalk? There are languages with no runtime, languages with some runtime, and languages with full image in place. Each has massively different pros and cons.

When you say Haskell and Smalltalk are name focused, you are technically right, but developer experience is extremely different.

I don't want to be overly negative, but it seems to me that author considers just different flavours of C.

There is a massive difference between Clojure, Prolog, and Forth.

The whole:

    type name = value—type-focused
    name: type = value—name-focused
    var name type = value—qualifier-focused

Is so much deep into details of how syntax might look like.

If you are choosing between Kotlin and Go, it is for the platform, not the syntax. If you decide between Haskell, Idris, Scheme, you do it with the syntax in mind.

All programming languages are basically Algol or Lisp, even the ones I like.

C? Basically Algol. Pascal? Basically Algol, actually quite closely. Go? Basically Algol, via Pascal. Lua? Basically Algol, surprisingly closely.

Forth? Basically Lisp. Postscript? Basically Lisp.

When I ponder about PL semantics, crossroad feature is currying. This decides whether partial application, pattern matching, function composition, combinators, and other functional features would play a major role in the language design. Functions are curried? ML style. Not curried? C style.

What's "ML" in this context?

ML (Meta Language)

https://en.wikipedia.org/wiki/ML_(programming_language)

Those syntax examples are literally just general examples, not the only way you have to write them. But to categorize the ones you wrote:

* Clojure is qualifier-focused * Prolog is name-focused * Forth is qualifier-focused (`:` is the qualifier in this case)

As Ken Iverson noted in "Notation as a Tool of Thought"[1], yeah the syntax absolutely matters. The same program might resonate and make sense in one language but be incomprehensible if translated 1:1 in another.

Computer languages are for humans to understand and communicate.

1. https://www.eecg.utoronto.ca/~jzhu/csc326/readings/iverson.p...

Iverson's point is more regarding semantics than syntax, though. The only mention of syntax suggests its better for it to be simple (presumably so that the semantics are closer to the surface). Every programming language is a notation for describing computation; notation is a catch all for all three levels: orthography, syntax, and semantics. APL is interesting because it not only uses an unconventional syntax, but also an unconventional orthography (obligate usage of special symbols), and its semantics are different as well from most languages (array programming). Iverson's point is that APL as a notation is valuable for making the structure of certain computations obvious, and that this point generalizes across programming languages.

GingerBill's article is making a narrower claim: that semantics are what determines a good notation usually, not syntax.

> Iverson's point is that APL as a notation is valuable for making the structure of certain computations obvious,

Yes, very much so, where as usual, there are local maxima where each notation excels at representing some problems nicely and is awkward for others.

A languages syntax and its error messages are its user interface. Yes you can have a good tool that you don’t enjoy looking at. You can also have a good tool that’s frustrating to learn because its user interface isn’t clear and doesn’t do what you expect. Can I not hope for something that does what I need, is easy to use, and looks good?

I dislike the “you can change the syntax” argument because that just doesn’t happen. Closest thing is a new language that compiles to another.

> Lua is an example of such a language, and when a semicolon is necessary is when you have something that could be misconstrued as being a call:

    (function() print("Test1") end)(); -- That semicolon is required
    (function() print("Test2") end)()

Tangential, but I sidestepped this ambiguity in a language I've been designing on the side, via the simple rule that the function being called and the opening parenthesis can't have whitespace between them (e.g. "f()" is fine but "f ()" or "f\n()" is not). Ditto for indexing ("x[y]"). If these characters are encountered after whitespace, the parser considers it the beginning of a new expression.

By sacrificing this (mostly unused, in practice) syntactic flexibility, I ended up not needing any sort of "semicolon insertion" logic - we just parse expressions greedily until they're "done" (i.e. until the upcoming token is not an operator).

I definitely think needless whitespace flexibility often causes problems. For eg I'm pretty sure Bjarne chose :: instead of : for the namespace operator in C++ due to ambiguity. A little bit of required whitespace around jump labels and ternary expressions and we could have saved an extra character in an operator that often occurs multiple times per line. Everybody runs linters that enforce that anyways. Likewise the inability to use a hyphen in an identifier has wasted a lot of my time over the years, but nobody uses squashed subtraction expressions.

> nobody uses squashed subtraction expressions

I do.

That sidestepping only introduces significant whitespace, which may not be what people want in the first place. It can be quite common to align call expressions by the parentheses if they span multiple lines for clarity, and you preventing might not be desireable.

I wouldn't like that, because I always put a single space in front of the parenthesis. Mostly for aesthetic reasons, and to distinguish it from function-like macros. Not intended as a criticism, just a data point.

I personally would prefer to hear more about what is uniquely good about Odin semantically or syntactically than more ad hominem attacks on the intelligence of the critics of the language, which I have seen in multiple recent pieces by this author.

Agreed; this communication style by itself makes me less inclined to try out the language.

in the era of LLMs, syntax might matter more than you think.

The c form of `type name;` is ambiguous because it could actually be more than one thing depending on context. Even worse if you include macro sheananigans. The alternate (~rust/zig) is `var/const/mut name type` is unambiguous.

For humans, with rather long memory of what is going on in the codebase, this is ~"not a problem" for experts. But for an LLM, its knowledge is limited to the content that currently exists in your context, and conventions baked in with the training corpus, this matters. Of course it is ALSO a problem for humans if they are first looking at a codebase, and if the types are unusual.

I hope that someday LLMs will interact with code mostly via language servers, rather than reading the code itself (which both frequently confuses the LLM, as you've noted, but is also simply a waste of tokens).

why? I suspect that writing code itself is extremely token efficient (unless like your keywords happen to be silly, super-long alien text).

Like which do you think is more token-efficient?

     <tool-call write_code "my_function(my_variable)"/>

    <tool-call available_functions/>

    resp: 
         <option> my_function </option>
         <option> your_function </option>
         <option> some_other_function </option>
         <option> kernel_function1 </option>
         <option> kernel_function2 </option>
         <option> imported_function1 </option>
         <option> imported_function2 </option>
         <option> ... </option>
     <tool-call write_function_call "my_function"/>
     resp:
         <option> my_variable </option>
         <option> other_variable_of_same_type </option>
     <tool-call write_variable "my_variable"/>

Not sure I follow. You seem to have omitted the part of 1) explaining how the LLM knew that my_function even existed - presumably, it read the entire file to discover that, which is way more input tokens than your hypothetical available_functions response.

Reading files is not that input tokien heavy, I suspect. But anyways I omitted it because presumably it would have done so to gain local context in general.

I still don't follow. Why doesn't that count as cost?

A+B > A+C => B > C

The LLM doesn't need to read the whole file in example 2). Why would it?

Where would it inject the code?

Presumably, the same MCP tool allowing it to read specific definitions would also allow it to write out specific definitions (rather than, again, having to write out the whole file just to make one change). Otherwise it defeats the purpose.

This is the same way humans work on code. You never read an entire file top to bottom just to find out one thing, and you never type out the entire file just to make one change. This would, understandably, be very taxing on your working memory, and you'd probably make more mistakes.

Honestly claude doesn't typically do this the way you think. It's perfectly capable of working in files that are "way too big for its allowed download window".

But in general you want some amount of context (what functions are available in scope, etc.)

LSP is meant for IDEs and very deterministic calls. Its APIs are like this: give me a definition of <file> <row> <column> <lenght>. This makes sense for IDEs because all of those can be deterministically captures based of your cursor position.

LLMs are notoriously bad at counting.

I think one could easily build an MCP tool wrapping LSP which smooths over those difficulties. What the LLM needs is just a structured way to say "perform this code change" and a structured way to ask things like "what's the definition of this function?" or "what functions are defined in this module?"

Not much different from what agents already do today inside of their harnesses, just without the part where they have to read entire files to find the definition of one thing.

So not using LSP, but rather using something in a middle that uses LSP as implementation detail.

So far enabling LSP in Claude only added messages like "this is old diagnostic before my edit".

Humans also have limited context. For LLMs it's mostly a question of pipeline engineering to pack the context and system prompt with the most relevant information, and allow tool use to properly understand the rest of the codebase. If done well I think they shouldn't have this particular issue. Current AI coding tools are mostly huge amounts of this pipeline innovation.

I think we need a LLM equivalent of this part's of fitt's law: The fastest place to click under a cursor is the location of the cursor. For an LLM the least context-expensive feedback is no feedback at all, the LLM should be able to intuit the correct code in-place, at token generation.

I suspect that the context-dependence in C is more an issue with the implementation than the overall syntactic philosophy.

It's a thing in c.

foo * bar;

Is that multiplication? Or a declaration of type foo*?

Right; and my argument is that this isn't because the type expression `foo *` precedes the name `bar`; it's because the type "pointer to foo" is expressed in a way that could also be a prefix of a multiplication expression.

Yes, it's a pity that some symbols have been overloaded with multiple meanings and the syntax can't disambiguate them. This case would not exist if another symbol had been chosen for dereference.

A similar claim, equally not related to type declarations, is with & that serves two roles as a boolean operator and address-of, that makes (foo)&bar ambiguous.

The list of C symbols that have been badly chosen at the beginning and now it's too late to correct continues with && for logical and, that prevents masking an address-of without adding spaces or parentheses in ``mask & &var``, /* for comments, that prevents dividing by a dereferenced value without adding spaces or parentheses in ``*a / *b``, the pre-increment ``--`` that is indistinguishable from a double application of the unary minus, and the original =+ etc. that have been later changed to += etc.

Ok fine but never here have I not been talking specifically about C

Major syntactic structures definitely have an influence on my language choices. Outside of compilation and runtime model, modeling the domain (both data and procedures) changes drastically between paradigms. Syntax is what enables or hamstrings different modeling paradigms.

My two biggest considerations when picking a language are:

- How well does it support value semantics? (Can I pass data around as data and know that it is owned by the declaring scope, or am I chained to references, potential nulls, and mutable handles with lifetimes I must consider? Can I write expression-oriented code?)

- How well does it support linear pipelining for immutable values? (If I want to take advantage of value semantics, there needs to be a way to express a series of computations on a piece of data in-order, with no strange exceptions because one procedure or another is a compiler-magic symbol that can't be mapped, reduced, filtered, etc. In other words, piping operators or Universal Function Call Syntax.)

I lean on value semantics, expression-oriented code, and pipelining to express lots of complex computations in a readable and maintainable manner, and if a language shoots me in the foot there, it's demoralizing.

Syntax, or how humans perceive the syntax, is only a very small part of the problems when designing a programming language. There is a lot more about how a compiler would handle the syntax (efficiently) and about how the syntax affects actual code and ecosystem.

The recent go blog on error handling should make it clear that syntax is often not worth worrying about. https://go.dev/blog/error-syntax

A lot of programming is taste, and syntax gives you a very quick judgement about how good the language designer's taste is. How familiar they are with what we know about which syntax works well, and so on. For example if you're designing a language in 2026 that uses `type name` instead of `name: type`... that is highly suspicious.

Also syntax is the interface through which you interact with the language, so bad syntax is going to be something annoying you have to deal with constantly. Sure you'll be able to write good programs in a language with bad syntax choices, but it's going to be less fun.

> Odin’s rules, which are very similar to Python’s, are to ignore newline-based “semicolons” within brackets (( ) and [ ], and { } used as an expression or record block).

Honestly I always thought that was a bit crap in Python and I'm surprised anyone thought this was a sensible thing to copy. Really, just use semicolons. As soon as an "easy" rule becomes even vaguely difficult to remember it's better to bin it and just require explicitness, because overall that is easier.

Did you even read the article, even the heading of that section? You know there is a huge majority that do not want semicolons whatsoever, so saying people should just suck it up is actually a determent to the adoption of the language because of people's general tastes.

This is one of those things that I’m glad remains highly personal and stochastic. Programming languages thrive when people are able and willing to create value with them. People thrive when there are options and they can find a language that sings to them.

Hmm... I don't know how serious this wants to be taken

- first example argues that syntax is interchangeable

- but then the author tries to sell me his semis

???

And who even cares in the time of LLMs and auto formatting?

"My hammer has this slight angle and if I hit the nail with just the right motion, it feels really good"

I was trying to be EXTREMELY clear I am talking specifically about DECLARATION syntax only and how for the most part, the concrete syntax for the DECLARTAION syntax is not that important in many languages.

However concrete syntax matters a heck of a lot and cannot be trivially interchangeable.

I had to write a follow-up article to clarify this because of people like yourself wrongly interpreting the article: https://www.gingerbill.org/article/2026/02/21/does-syntax-ma...

I like the semantics you type in the google search bar when using it for impromptu calculations. You can use ^ to raise to a power, for example. Just type sin 45. It’s all least surprise.

I love how it works with units too, eg. c/433MHz or 4 bytes * 20hz * 24 hours

Syntactically it probably has a ceiling, to be so casual. Least surprise won’t work for very complex programs. But maybe the programs wouldn’t be so complex if you didn’t have to stick together complex program syntax either.

I wonder how many people use REPL of a language of choice as a calculator as I do.

Python doesn't have automatic semicolon insertion.

Yes it does. You can even type `;` anywhere too, go an try it.

Because of how good the rules are for Python's needs, most people never know that semicolons even exist in Python.

This isn't quite right. Go and JavaScript use semicolons pervasively throughout their formal grammars, and mostly don't use newlines; then, on top of that, they have special rules that allow the semicolons to be omitted in many cases depending on newline placement. Python uses newlines pervasively throughout its formal grammar, and only allows semicolons to be used in a couple specific places (at the end of a simple statement or to separate multiple simple statements on a single line).

The two systems have in common that you can either include or omit a semicolon when you've got a simple statement on its own line, but under the hood they're different; the missing semicolon at the end of a line in idiomatic Python is not "automatically inserted by the parser", it is simply not needed at all. There are also behavioral differences; for example, in Go and JavaScript a semicolon on a line by itself is legal and does nothing, while in Python it's a syntax error.

This might might sound weird but I think that's a distinction without a difference.

Those "newline tokens" are effectively the equivalent to a semicolon in the parser. They are statement terminators. The way that Odin or Python handles them is effectively the same except when I tokenize the newlines, they are treated as semicolons but with the "text" being "\n". Thus the distinction here is a matter of what you call it, not how it is treated.

That is the same as a form of "automatic statement-terminator insertion" where that statement-terminator is whatever you want it to be. And to me, that is as form of "semicolon" (which is being used as a generalization for statement-terminator) insertion.

Python actually has optional semicolons.

I never got why compilers don't have pluggable syntaxes.

I mean, once you decide the "flavor" (e.g.: typed, imperative, with a dash of functional and some oop for good measure), you could have more than one syntax and easily switch to whatever the reader wants.

We had an integration language in a product I worked on that had three flavors (you can check it here: https://docs.oracle.com/cd/E13154_01/bpm/docs65/pdf/OracleBP... , page 254)

The original syntax scared some people, so we had the compiler use the same AST with three different parsers: Original, Java and VB. The editor (which had syntax highlighting and auto completion) would let you see the code however you wanted.

You could even have a setting in the IDE that always showed the code as you wanted.

We even respected some weirdness in the spacing and indentation of comments and code when needed.

For some languages, like rust it may be a stretch, but for most vanilla languages, you could easily re-skin them to look much more like something else, that's comfy for whoever is looking at the code.

> I never got why compilers don't have pluggable syntaxes.

An interesting question, but the answer is "because it's a bad idea" that doesn't actually solve the problem.

That said, the right way to implement this is as a "transpiler" that compiles one syntax into another. And only the people who want to use it pay the costs.

> An interesting question, but the answer is "because it's a bad idea" that doesn't actually solve the problem.

This doesn't really explain anything, and it isn't clear that both of you have the same model of "the problem" in mind.

The usual way people get here is that they didn't realise programs are for reading by humans. That's why we have formatting conventions (the compiler doesn't care but humans do) and so it's also why a single syntax is important.

If there are six syntaxes for a hypothetical language L, then either every L practitioner must learn all six syntaxes (ew, no thanks) or most L programmers can't read each other's programs and so it's basically unmaintainable.

Code is communication. The compiler could handle it, but what is important is that other people can.

There are many infamous examples of people using the C preprocessor to write near-Pascal or similar in C. It largely died out because it hindered effective communication about the code.

> I never got why compilers don't have pluggable syntaxes.

Because then you've created dialects and produced the Curse of Common Lisp all over again.

Syntax matters a heck of a lot, and you want consistency and coherency across codebases of the same language, otherwise you don't have a language at all. So wanting it "pluggable" is in fact the worse possible choice you want.

It's also a naïve view to think that concrete syntax can be trivially swapped out with the abstract syntax remaining. For certain things it can usually work (like declaration syntax in the article) but for most things it cannot.

> I am still perplexed by how people judge a language purely by its declaration syntax, and will decide whether to use the language purely based on whether they like that aspect or not.

Throughout the article, OP seems baffled that people have aesthetic preferences. Well, yes, of course we do; dealing with ugly things is the computer's job.

It also comes across like OP hasn't seen a lot of examples of really interesting language syntax, i.e. things outside, shall we say, the extended Algol family. The discussion seems to accommodate brace-less languages like Python, but not e.g. the Lisp or Forth families.

> and thus just becomes a question of ergonomics or “optimizing for typing” (which is never the bottleneck).

It might not be a bottleneck in terms of time needed. But unpleasant syntax is annoying and breaks flow. Thoughts creep in about how you wish the language looked different and that you didn't have to type these other bits. (Which is why a lot of people feel so strongly about type inference.)

> From what I gather, this sentiment of not understanding why many “modern” languages still use semicolons is either:

OP seems to conflate "semicolon" with "visible, explicit token that separates statements. There's no reason it couldn't be some other punctuation, after all. Describing Python's approach to parsing as "automatic semicolon insertion" is wild to me; the indented-block structure is the point and statements are full lines by default. Semicolons in Python are a way to break that rule (along with parentheses, as noted), which are rarely seen as useful anyway (especially given the multiple assignment syntax).

> To allow for things like Allman braces, Odin allows for extra single newline in many places in its grammar, but only an extra single newline. This is to get around certain ambiguities between declaration a procedure type and a procedure literal

Right; and the point of Python's approach is to not need braces in the first place, and therefore sidestep any considerations of brace style. And when you do that, it turns out that you don't need to think nearly as hard about whether a newline should terminate a statement. It's a package deal.

> Maybe I don’t need to be as cynical and it is a lot simpler than all of that: first exposure bias. It’s the tendency for an individual to develop a preference simply because they became familiar with it first, rather that it be a rational choice from a plethora of options.

> However I do think there are rational reasons people do not like a syntax of a language and thus do not use it. Sometimes that syntax is just too incoherent or inconsistent with the semantics of the language. Sometimes it is just too dense and full or sigils, making it very hard to scan and find the patterns within the code.

For what it's worth, before I ever touched Python I had already used (in no particular order) multiple flavours of BASIC, Turing, probably at least two kinds of assembly, Scheme, C, C++, Java and Perl. To be fair, I had also used HyperTalk and Applescript, so maybe that does explain why I glommed onto Python. But BASIC came first.

In my mind, a mid-line semicolon is exactly the kind of sigil described here, and an end-of-line sigil is simply redundant. Multi-line statements should be the explicitly-marked exception, if only because long statements should be less common than shorter ones.

> Throughout the article, OP seems baffled that people have aesthetic preferences. Well, yes, of course we do

I think Bill is astonished not really that people have aesthetic preference but that anybody's real preferences could be different from his. He supposes that, at most, these are false beliefs they should reject to reveal that their sincere preference is identical to his, the only correct preference.

Hello again. That's not my position in the slightest. The point of the article is to express how people will shallowly and naïvely judge something. It might be naïve conception of "aesthetics" but it's not even a good one.

It has nothing to do with people disagreeing with me, I have seen people dismiss numerous languages (not just my own) based on that declaration syntax. Things like "why does Rust not use `type name = value`? why did they have to change things so much from C++?".

Do you think these people are actually serious programmers or just having shallow and dumb opinions on things they know little to nothing about?

> Do you think these people are actually serious programmers [...] ?

I can't possibly tell? The surface aesthetic impression is only that.

I tried my best to be very clear within this article that I am talking about two specific things: DECLARATION SYNTAX and SEMICOLONS. This does not extend to anything else. In fact I got so frustrated with people like yourself who are wrongly interpreting the article that way and extrapolating too much that I had to write a new article: https://www.gingerbill.org/article/2026/02/21/does-syntax-ma...

"Does Syntax Matter?" And the first one is "Yes."

And all languages can have their declaration syntaxes categorized into one of the three I mention. This is not limited to ALGOLs. And I have come across "really interesting language syntax". I am not talking about that in this article.

----

Ergonomics does matter and it's not just about optimizing for typing either. It's actually all about coherency and consistency, something I go into depth into the following article on syntax.

----

I am not "conflating" a semicolon with statement terminators, I am using that as shorthand as most languages that most people encounter (not all languages, obviously) will use semicolons. Python's approach to parsing is a form of automatic semicolon insertion, even if most people don't think of it that way.

----

And I do not like Python's approach requiring significant whitespace of "blocks" through indentation. I much prefer using something like braces, or an `end` keyword, etc, literally because it prevents the problems that significant whitespace introduces. I am not getting into that argument here; please don't start it.

Reply on news.ycombinator.com