| Commit message (Collapse) | Author |
|
|
|
Also, put Win32 to extra-deps (the then-unreleased version was
previously fetched in a .zip tarball from GitHub directly).
|
|
|
|
|
|
|
|
|
|
I don't know what's happening.
The latest string of commits is basically broken at every point.
I'll tag as soon as I decide it's safe to use, which probably will
become the default mode of development (instead of _every_ commit
representing a working version).
|
|
Yeah, so last night I added a Show instance to WindowsEnv.Value
(ex-WindowsEnv.VarValue), which would simply call valueString
(ex-varValueString).
Then I replaced every valueString/varValueString with show.
Then I decided that this Show instance was a mistake, and derived an
instance instead.
It obviously messed up all the show VarValue/Value calls.
The lesson to learn is you should remove an instance first, fix all the
call sites, and only then derive it.
The disgusting part is that a few of the last commits are actually
broken, which I hate.
|
|
|
|
|
|
Also, fix compiler warnings (I've got too used to building with
`--ghc-options -w`).
|
|
|
|
The fact whether the registry value was a regular or an expandable
string is now propagated up to the `Environment` module (and even
further to the apps).
This was done to get rid of these weird `setString*` functions (and the
like).
I don't feel like I've came up with the right abstractions yet though,
so there's more work on this to come.
|
|
|
|
|
|
|
|
|
|
|
|
|