But loading all the required files from an "all.lisp" file makes asdf unaware of the existence of those files, doesn't it? I can use that as a workaround but it has several disadvantages: The relationships between those files and the rest of the system will be lost, and any changes with regards to compilation/loading configuration (bundling, loading from pre-compiled files) will have to be done manually in all.lisp, hidden from the project .asd file.

Given that a workaround is necessary for this, using a reader macro in the system definition to unfold a list of :file components from a glob pattern seems like a better solution, but it's a shame that such a basic feature isn't supported out of the box.

In my opinion, if you really want to push asdf as the standard system definition for Common Lisp, these kind of things (what the users need, after all) should be given more attention. It's unusual for a programming ecosytem to have a powerful system definition tool and still having to rely on Makefiles and roll-your-own hacks for some things.

Don't get me wrong, I think asdf is a great tool and I thank you all for the effort you put on it.

Kind regards, rcn

wild-modules were an example extension and/or contrib from the asdf 1 days. There is a test for them in test/ but otherwise I don't believe anyone uses them. If you have the asdf git installed, you can use them with package-inferred-system.

I believe that package-inferred-system is a better way to achieve a similar effect, that allows for dependencies between sub-systems, unlike wild-modules. I suppose a wildcard option to load all files in a directory while using package-inferred-system dependencies would be a nice feature to add. But so far, I found it more practical to have an explicit file named "all.lisp" that includes all the files that matter.