On Sat, Jul 16, 2022 at 12:22:49AM -0400, Brad Smith wrote: > On 7/15/2022 11:59 PM, Willy Tarreau wrote: > > Hello, > > > > On Fri, Jul 15, 2022 at 07:27:12PM -0400, Brad Smith wrote: > > > On 7/15/2022 1:34 AM, ???? ??????? wrote: > > > > I wonder how do NetBSD/OpenBSD ports work, do they use their own > > > > "install" invocation instead of "make install" ? > > > > shouldn't they switch to "make install" ? > > > NetBSD uses the Makefile's install targets but patches out the -v flag. > > > OpenBSD used it's own install target > > > but I'm trying to remove that special casing and I had basically the same > > > sort of diff NetBSD has. Can't switch > > > to the make install target until the Makfile is fixed. > > I agree with the principle of your patch, just not with the way it's done, > > because this variable "IV" is a bit cryptic and not easy to follow. Other > > programs also like to redefine the install program, thus I'd propose a > > simpler and more flexible approach: > > > > - define "INSTALL = install" early, next to DESTDIR and friends > > - set "INSTALL = install -v" in the LINUX targets > > - use $(INSTALL) in the install targets > > > > This way it even allows users of any platform to simply pass the INSTALL > > variable to match their needs (including setting it to "ginstall -v" on > > non-linux platforms where this often points to GNU install). > > > Hi Willy, > > I kind of figured you would not be Ok with it as is. I did think about doing > something > as you suggested today as I was thinking about this. I was trying to keep > the > variable name short and I agree with you. What you have suggested is typical > in autoconf / automake environments.
Yes, INSTALL is among the common variables to look for in makefiles. Willy
