[jasonpeacock]

C libraries have advertised "header-only" for a long time, it's because there is no package manager/dependency management so you're literally copying all your dependencies into your project.

This is also why everyone implements their own (buggy) linked-list implementations, etc.

And header-only is more efficient to include and build with than header+source.

[uecker]

I never copied my dependencies into my C project, nor does it usually take more than a couple of seconds to add one.

[AlotOfReading]

There's a number of extremely shitty vendor toolchain/IDE combos out there that make adding and managing dependencies unnecessarily painful. Things like only allowing one project to be open at a time, or compiler flags needing to be manually copied to each target.

Now that I'm thinking about it, CMake also isn't particularly good at this the way most people use it.

[uecker]

They are certainly bad vendor toolchain, but I want to push back against the idea that this is a general C problem. But even for the worst toolchains I have seen, dropping in a pair of .c/.h would not have been difficult. So it is still difficult to see how a header-only library makes a lot of sense.

[AlotOfReading]

One of the worst I've experienced had a bug where adding too many files would cause intermittent errors. The people affected resorted to header-izing things. Was an off-by-one in how it was constructing arguments to subshells, causing characters to occasionally drop.

But, more commonly I've seen that it's just easier to not need to add C files at all. Add a single include path and you can avoid the annoyances of vendoring dependencies, tracking upstream updates, handling separate linkage, object files, output paths, ABIs, and all the rest. Something like Cargo does all of this for you, which is why people prefer it to calling rustc directly.

[uecker]

People certainly sometimes create a horrible mess. I just do not see that this is a good reason to dumb everything down. With a proper .c/.h split there are many advantages, and in the worst case you could still design it in a way that it is possible "#include" the .c file.

I tried to use cargo in the past and found it very bad compared to apt / apt-get (even when ignoring that it is a supply-chain disaster), essentially the same mess as npm or pip. Some python packages certainly wasted far more time of my life than all dependencies for C projects I ever had deal with combined.

[AlotOfReading]

I consider it compiler abuse to #include a source file. Useful for IOCCC competitions though.

Apt is fine for local development, but it's a bit of a disaster for portability and reproducibility. Not uncommon to see a project where the dependencies either have unspecified versions whose latest versions in universe are incompatible, or where the package has been renamed and so you have to search around to find the new name. Plus package name conventions are terrible. libbenchmark-dev, libgoogle-perftools-dev, and libgtest-dev are all library packages from the same company. The second one is renamed to gperftools-lib with RPM repos, to further compound the inconsistency.

I find myself dealing with package and versioning rename issues regularly in the CI pipelines I have using apt.

[kazinator]

> And header-only is more efficient to include and build with than header+source.

Dispute.

This is C code. You can't just drop it in and build it; you have to write code to use it. You have to figure out the API to correctly use it. If memory is passed around by pointers, you have to understand the responsibilities: who allocates, who frees, who may touch what when.

In the first place, you have to decide whether to commit to that library that far; it might not be until you've done some exploratory programming with it that you want to scrap it and find another one.

The cost of adding two files versus one is practically nothing in consideration of the larger picture.

The separate header model is baked into the C mindset; the tooling readily supports it.

Many "header only" libraries are secretly two files in one. When you define a certain preprocessor symbol before including the file, you get the implementation bits. When you don't define that symbol, the header is a pure header.

That means you have to pick some existing .c file which will define the preprocessor symbol and include the heaader. That source file becomes a "surrogate" source file, taking the place of the source file it ought to have.