On Thu, Jun 20, 2024, at 17:34, Shubham Kanodia wrote: > I've seen libraries in a few language ecosystems use `ini` parsers for > parsing the config file, and there's several blogs (that perhaps > incorrectly?) state so, so > I assumed that might be true. But you're right in that I don't see it > be mentioned on git's official site. I see no warrant for such an assumption. In my experience, plenty of syntaxes are described as like-X”. Java and C# are C-like. That doesn’t mean you can use a C parser on those other languages. Config files are simpler but the same principle applies. > 1. What spec does the config file follow? Apparently there isn’t a spec because it is bespoke. https://stackoverflow.com/a/68461700/1725151 > 2. What is the correct way then to get an "effective" git config > value? Typically, I assumed that if a value appeared twice in the git > config, the second would override the first (for say, `core.editor`). > How does git parse "overrides" vs "arrays" if they are defined using > the same syntax? There are two dimensions 1. How config variables are parsed 2. What is expected of the specific config variable `core.editor` is a single value. You can test with ``` [core] editor=vim editor=nano ``` The last one wins here. `core.editor` expects a single value. But you can define a multi-valued variable ``` [customsection] mycustomvariable = value1 mycustomvariable = value2 mycustomvariable = value3 ``` ``` $ git config --global --get-all customsection.mycustomvariable value1 value2 value3 ``` -- Kristoffer Haugsbakk
