| Message ID | 20220120192151.k46VF%steffen@sdaoden.eu (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Makefile: uses rsync(1), could this be optional? | expand |
On Fri, Jan 21, 2022 at 4:31 AM Steffen Nurpmeso <steffen@sdaoden.eu> wrote: > > Hello. > > I sent this to linux-kernel@vger.kernel.org on the 15th, which > seems to be legacy. Just in case someone is wondering about the > resend. I did not see your previous post. What is bad about using rsync? > > As a not-yet-tested low-quality Makefile suggestion, with modern > GNU tools and find(1)'s -printf, wouldn't the following code work > out gracefully in practice? (Not subscribed.) > > Thanks for Linux! > > --- Makefile.orig 2022-01-15 19:33:59.337393371 +0100 > +++ Makefile 2022-01-15 19:34:07.447393217 +0100 > @@ -1260,8 +1288,17 @@ export INSTALL_HDR_PATH = $(objtree)/usr > quiet_cmd_headers_install = INSTALL $(INSTALL_HDR_PATH)/include > cmd_headers_install = \ > mkdir -p $(INSTALL_HDR_PATH); \ > - rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ > - usr/include $(INSTALL_HDR_PATH) > + if command -v rsync; then \ > + rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ > + usr/include $(INSTALL_HDR_PATH);\ > + else \ > + cd usr;\ > + find include/ -type f -name '*.h' -printf '%f %h\n' |\ > + while read f d; do \ > + mkdir -p $(INSTALL_HDR_PATH)/$$d;\ > + cp -P $$d/$$f $(INSTALL_HDR_PATH)/$$d/$$f;\ > + done;\ > + fi > > PHONY += headers_install > headers_install: headers > > --steffen > | > |Der Kragenbaer, The moon bear, > |der holt sich munter he cheerfully and one by one > |einen nach dem anderen runter wa.ks himself off > |(By Robert Gernhardt) -- Best Regards Masahiro Yamada
Guten Tag. Masahiro Yamada wrote in <CAK7LNAQOm8NYiTDQnd0P-UsGa7GurffQiWQgGh0Cze4wLmDmgA@mail.gmail.com>: |On Fri, Jan 21, 2022 at 4:31 AM Steffen Nurpmeso <steffen@sdaoden.eu> \ |wrote: |> I sent this to linux-kernel@vger.kernel.org on the 15th, which |> seems to be legacy. Just in case someone is wondering about the |> resend. | |I did not see your previous post. |What is bad about using rsync? Oh really nothing, but this Linux distribution (CRUX) recreates Linux headers before the GNU LibC is build, and this is the only dependency of rsync around. And, unless i am mistaken, the other code path is more expensive but otherwise functionally equivalent? |> As a not-yet-tested low-quality Makefile suggestion, with modern |> GNU tools and find(1)'s -printf, wouldn't the following code work |> out gracefully in practice? (Not subscribed.) |> |> Thanks for Linux! |> |> --- Makefile.orig 2022-01-15 19:33:59.337393371 +0100 |> +++ Makefile 2022-01-15 19:34:07.447393217 +0100 |> @@ -1260,8 +1288,17 @@ export INSTALL_HDR_PATH = $(objtree)/usr |> quiet_cmd_headers_install = INSTALL $(INSTALL_HDR_PATH)/include |> cmd_headers_install = \ |> mkdir -p $(INSTALL_HDR_PATH); \ |> - rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ |> - usr/include $(INSTALL_HDR_PATH) |> + if command -v rsync; then \ |> + rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ |> \ |> + usr/include $(INSTALL_HDR_PATH);\ |> + else \ |> + cd usr;\ |> + find include/ -type f -name '*.h' -printf '%f %h\n' |\ |> + while read f d; do \ |> + mkdir -p $(INSTALL_HDR_PATH)/$$d;\ |> + cp -P $$d/$$f $(INSTALL_HDR_PATH)/$$d/$$f;\ |> + done;\ |> + fi |> |> PHONY += headers_install |> headers_install: headers --steffen | |Der Kragenbaer, The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
On 1/21/22 08:36, Steffen Nurpmeso wrote: > Guten Tag. > > Masahiro Yamada wrote in > <CAK7LNAQOm8NYiTDQnd0P-UsGa7GurffQiWQgGh0Cze4wLmDmgA@mail.gmail.com>: > |On Fri, Jan 21, 2022 at 4:31 AM Steffen Nurpmeso <steffen@sdaoden.eu> \ > |wrote: > |> I sent this to linux-kernel@vger.kernel.org on the 15th, which > |> seems to be legacy. Just in case someone is wondering about the > |> resend. > | > |I did not see your previous post. > |What is bad about using rsync? > > Oh really nothing, but this Linux distribution (CRUX) recreates > Linux headers before the GNU LibC is build, and this is the only > dependency of rsync around. And, unless i am mistaken, the other > code path is more expensive but otherwise functionally equivalent? Do we need to add it to Documentation/Changes? > |> As a not-yet-tested low-quality Makefile suggestion, with modern > |> GNU tools and find(1)'s -printf, wouldn't the following code work > |> out gracefully in practice? (Not subscribed.)
--- Makefile.orig 2022-01-15 19:33:59.337393371 +0100 +++ Makefile 2022-01-15 19:34:07.447393217 +0100 @@ -1260,8 +1288,17 @@ export INSTALL_HDR_PATH = $(objtree)/usr quiet_cmd_headers_install = INSTALL $(INSTALL_HDR_PATH)/include cmd_headers_install = \ mkdir -p $(INSTALL_HDR_PATH); \ - rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ - usr/include $(INSTALL_HDR_PATH) + if command -v rsync; then \ + rsync -mrl --include='*/' --include='*\.h' --exclude='*' \ + usr/include $(INSTALL_HDR_PATH);\ + else \ + cd usr;\ + find include/ -type f -name '*.h' -printf '%f %h\n' |\ + while read f d; do \ + mkdir -p $(INSTALL_HDR_PATH)/$$d;\ + cp -P $$d/$$f $(INSTALL_HDR_PATH)/$$d/$$f;\ + done;\ + fi PHONY += headers_install headers_install: headers