When is safer a macro than user code --> Macro "cyclomatic" complexity

[pellico]

A macro could be safer than user code because human are very bad at executing repetitive tasks.

However a complex macro can be more dangerous than manual code when it is very complex and difficult to verify
When a macro is too dangerous ?
It could be help full to introduce a metrics to measure the risk of bugs in the macro --> "cyclomatic complexity index"

This topic is related to issue #71 . Maybe it is worth to merge.

[felix91gr]

Hey @pellico!

Maybe it is worth to merge.

I've just made this Issue be a Child Issue of #71, because it's not exactly a duplicate; I'd say it's more like a deep dive into one of its topics :)

It could be help full to introduce a metrics to measure the risk of bugs in the macro --> "cyclomatic complexity index"

I would like to be able to say "hell yeah" or "not worth it because of X", but as far as I'm aware, it's quite the nuanced topic. I went into some detail here.

Here is where I wish we had more prior literature on this kind of macros. I'll see if I can find something reasonable in an old Lisp forum; if I recall correctly, Rust's declarative macros are a kind of macro that was developed by the Lisp community a long time ago.

But long story short: measuring the maintainability / provability / understandability of code is either very hard or something we cannot yet do with our mathematical understanding. That being said, macros are not exactly code; they are something else. And this perhaps means that such limitations don't apply as strictly to them.

Remind me of this if I don't come back with the Lisp info during the following couple of weeks. I'll do my best to see what I can find of the macro analysis work that might've been done over there.