From: Phillip Wood Date: Mon, 16 Mar 2026 10:52:45 GMT Subject: Re: [PATCH 7/8] meson: compile compatibility sources separately Message-ID: <204db12f-daef-47c7-ab71-efd8c20ab10f@gmail.com> In-Reply-To: On 16/03/2026 08:09, Patrick Steinhardt wrote: > On Fri, Mar 13, 2026 at 10:33:20AM +0000, Phillip Wood wrote: >> >> It does mean we're using different includes when compiling with pre-compiled >> headers compared to compiling without them though which means contributors >> using per-compiled headers could accidentally depend on functions that are >> not included when compiling without them. Wasn't the idea behind >> "compat/posix.h" to avoid including "git-compat-util.h" in the reftable >> code? The commit message for 75a044f748f (git-compat-util.h: split out >> POSIX-emulating bits, 2025-02-18) says >> >> This intermixing is a bit of a problem for the reftable library as >> we don't want to recreate the POSIX-like interface there. But >> neither do we want to pull in the Git-specific functionality, as it >> is otherwise quite easy to start depending on the Git codebase >> again. >> >> We could precompile "compat/posix.h" for the code that does not want >> "git-compat-util.h" > > In theory, yes. But in practice we'd notice this quite fast via other CI > jobs, and it feels a bit ugly to split out so many different libraries. > So I'd propose to keep this as-is for now, but iterate in case we notice > that it _does_ become a problem. Fair enough, getting feedback via the CI when it could be given locally by the compiler isn't great but we don't expect this to be a common problem. Thanks Phillip > Does that work for you? > > Thanks! > > Patrick >