Incorrect HiGHS 1.13.1 with v1.14.2

[Thell]

I am not sure if this is a 'me', 'you' or 'them' issue but with the latest cargo crate version:

~\Workspaces\rust
❯ cd .\highs-sys-callback-test2\

Workspaces\rust\highs-sys-callback-test2
❯ cargo init
    Creating binary (application) package
note: see more `Cargo.toml` keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

highs-sys-callback-test2 on  master [?] is 📦 v0.1.0 via 🦀 v1.94.0
❯ cargo add highs-sys
    Updating crates.io index
      Adding highs-sys v1.14.2 to dependencies
             Features:
             + build
             + highs_release
             - discover
             - libz
             - ninja
    Updating crates.io index
     Locking 30 packages to latest Rust 1.94.0 compatible versions

highs-sys-callback-test2 on  master [?] is 📦 v0.1.0 via 🦀 v1.94.0
❯ explorer .

highs-sys-callback-test2 on  master [?] is 📦 v0.1.0 via 🦀 v1.94.0
❯ # copied main.rs

highs-sys-callback-test2 on  master [?] is 📦 v0.1.0 via 🦀 v1.94.0
❯ cargo run
   Compiling proc-macro2 v1.0.106
   Compiling unicode-ident v1.0.24
   Compiling glob v0.3.3
   Compiling quote v1.0.45
   Compiling libc v0.2.186
   Compiling prettyplease v0.2.37
   Compiling shlex v1.3.0
   Compiling memchr v2.8.0
   Compiling regex-syntax v0.8.10
   Compiling minimal-lexical v0.2.1
   Compiling windows-link v0.2.1
   Compiling find-msvc-tools v0.1.9
   Compiling either v1.16.0
   Compiling bindgen v0.72.1
   Compiling log v0.4.29
   Compiling rustc-hash v2.1.2
   Compiling libloading v0.8.9
   Compiling bitflags v2.11.1
   Compiling itertools v0.13.0
   Compiling clang-sys v1.8.1
   Compiling cc v1.2.62
   Compiling nom v7.1.3
   Compiling regex-automata v0.4.14
   Compiling syn v2.0.117
   Compiling cmake v0.1.58
   Compiling cexpr v0.6.0
   Compiling regex v1.12.3
   Compiling highs-sys v1.14.2
   Compiling highs-sys-callback-test2 v0.1.0 (C:\Users\thell\Workspaces\rust\highs-sys-callback-test2)
warning: unused variable: `message`
 --> src\main.rs:7:5
  |
7 |     message: *const i8,
  |     ^^^^^^^ help: if this is intentional, prefix it with an underscore: `_message`
  |
  = note: `#[warn(unused_variables)]` (part of `#[warn(unused)]`) on by default

warning: `highs-sys-callback-test2` (bin "highs-sys-callback-test2") generated 1 warning (run `cargo fix --bin "highs-sys-callback-test2" -p highs-sys-callback-test2` to apply 1 suggestion)
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 1m 00s
     Running `target\debug\highs-sys-callback-test2.exe`
Running HiGHS 1.13.1 (git hash: n/a): Copyright (c) 2026 under MIT licence terms

Note the Running HiGHS version is 1.13.1 yet highs-sys v1.14.2 was built. And when I check the build artifacts it is indeed 1.13.1 being built. So it looks like a version mismatch but looking at the git submodule version of the 1.14.2 tag it looks like the right hash 7df0786 so I am a bit mystified.

[lovasoa]

Hi ! Can you explain your issue ? Is this just that you find the version numbering system between the highs crate, the highs-sys crate, and the upstream highs project confusing ? Or is there a real issue or limitation you encountered ?

[Thell]

Hi ! Can you explain your issue ? Is this just that you find the version numbering system between the highs crate, the highs-sys crate, and the upstream highs project confusing ? Or is there a real issue or limitation you encountered ?

Well, if it is meant to execute HiGHS 1.13.1 instead of HiGHS 1.14 when the highs-sys version is 1.14.2 and the submodule version points to HiGHS v1.14.0, the _yes, I find that confusing. :)

Looking the highs-sys commits then:

8f13b1a

is obviously updating the HiGHS library, and I could only assume

7279196

(which updates the submodule) updates the HiGHS library also, this time to 1.14

Yes? And, as you can see from the output in the original post HiGHS 1.13.1 is being executed.

[Thell]

I wanted to let you know that I cloned the repo, and ran the tests (after adding version output) it shows HiGHS v1.1.4.0. So that part is good.

Perhaps it is an issue with what crates.io is pointing to? I tried to put the repo as a dependency directly but because of the submodule that didn't work so well (https -vs- ssh).

edit: Yep... that seems to be the difference... according to doc.rs the crates.io source is still version 1.13.1: (https://docs.rs/crate/highs-sys/1.14.2/source/HiGHS/Version.txt) but the repo shows 1.14.0

[Thell]

Just a final follow up for you...

highs-sys = { version = "1.14.2" }  # Results in HiGHS 1.13.1 executing
highs-sys = { version = "=1.14.1" } # Results in HiGHS 1.14 executing

[lovasoa]

See https://doc.rust-lang.org/cargo/reference/resolver.html

To know which version highs-sys = { version = "1.14.2" } resolves to in your project, you can look at your Cargo.lock

[Thell]

See https://doc.rust-lang.org/cargo/reference/resolver.html

To know which version highs-sys = { version = "1.14.2" } resolves to in your project, you can look at your Cargo.lock

Perhaps I wasn't being clear. It is the HiGHS that the highs-sys builds that is wrong. Both highs-sys versions resolve appropriately to the version entered. Meaning '1.14.2' resolves to '1.14.2' and '=1.14.1' resolves to '=1.14.1'.

The problem is the source that crates.io has for the latest 'highs-sys 1.14.2' release is 'HiGHS 1.13.1' instead of 'HiGHS 1.14.0' which is what the source for 'highs-sys 1.14.1' on crates.io shows. As was shown in the earlier post linking to the source listing on crates.io

Using either dependency (highs-sys = "=1.14.1" or highs-sys = "1.14.2") reveals the issue clearly on my end (Windows and Ubuntu WSL)

use std::ffi::CStr;

use highs_sys::*;

fn main() {
    unsafe {
        println!("HiGHS version: {:?}", CStr::from_ptr(Highs_version()));
    }
}

It isn't a big issue for me as I was just going to test some callback handling and can simply pin the highs-sys to 1.14.1 but it does seem like something that should be fixed ( by re-publishing to crates.io ? )

[jajhall]

Note that HiGHS v1.15 will be released soon...

[Thell]

Note that HiGHS v1.15 will be released soon...

From the recent latest branch commits...

The highlights of v1.15 are the first variant of the parallel MIP solver

hehe, you likely know how much I have been wanting this! Perhaps I will simply wait on my callback testing since the point of that was to setup a first-to-finish multi-process controller in Rust similar to the Python multi-process solution sharing controller.

Color me excited.

@lovasoa If you want to simply close this since the v1.15 publishing will likely resolve the issue on the crates.io side of things that'd be fine but if you do leave it open until you find out what happened I'd be interested in understanding how the source reverted back to HiGHS 1.13.1 without the git submodule hash changing.