Joel Rosdahl

Try this: ```emacs-lisp (use-package clang-format+ :hook (c-mode-common . clang-format+-mode)) ```

Looks like @Mingli-Yu forgot to mention that the issue [only shows up when using `-Og`](https://github.com/ccache/ccache/issues/1085#issuecomment-1149026113).

Looks like a solid implementation. However, I'm not yet sure that we actually should support HTTPS in this way, so I would like to discuss our options first. See #894.

> That sounds silly. Why would an automatic clean-up honour such a useful option, but not a manual clean-up? In fact, [the `limit_multiple` option was added](https://github.com/ccache/ccache/commit/a8ca6ad32c338cf145826e8091c5e3aeeb7f538a) because `ccache -c` **did**...

> The only reason to do a manual clean-up would then be to prevent performance hits due to automatic clean-ups at inconvenient times. Or am I missing something? Yes, there...

Yes, support for ISPC could be accepted if it's not too intrusive or complex.

> With these changes I made some adaptations that the tests run under Windows again. Note that the Windows tests are work in progress and have never worked before, so...

@rvogg wrote: > I thought that because there were many if ! $HOST_OS_WINDOWS are spread over the tests I understand. @nickhutchinson made several Windows-related improvements (#757, #780, #789, maybe more)...

@rvogg wrote: > I think mocking is a good idea. But why do you need your own execution API for that? Since ccache calls an external program anyway you could...

I don't consider this to be a bug since because the behavior is documented. Any improvements in the area are of course welcome. @vaneyli wrote: > But the __has_include is...