Issue #8115 has been updated by vo.x (Vit Ondruch).

ruby -v set to ruby 2.0.0p247 (2013-06-27 revision 41674) [x86_64-linux]

usa (Usaku NAKAMURA) wrote:
> On Windows, when making extensions, `nmake install' now causes error because rubylibprefix (and others) has "C:C:" prefix.
> I think we need to split ruby's path and extensions one more strictly.

I would say that r39841 should be reverted, since it never fixed my issues. Moreover, it introduces another issue, e.g. from my makefile:

vendorarchdir = $(DESTDIR)/usr/lib64/ruby/vendor_ruby
RUBYARCHDIR   = $(DESTDIR)$(vendorarchdir)$(target_prefix)

i.e. $(DESTDIR) is applied 2 times.

Not sure how I tested my patch, though. Sorry :/
----------------------------------------
Bug #8115: make install DESTDIR=/my/install/path fails
https://bugs.ruby-lang.org/issues/8115#change-40255

Author: vo.x (Vit Ondruch)
Status: Assigned
Priority: Normal
Assignee: nobu (Nobuyoshi Nakada)
Category: 
Target version: 
ruby -v: ruby 2.0.0p247 (2013-06-27 revision 41674) [x86_64-linux]
Backport: 


=begin
In Fedora packaging, we used to use (({make install DESTDIR=/my/install/path})) to install compiled extension into the folder, which RPM used to pick up the extension. However, since Ruby 2.0.0, this approach fails with message such as;

 make install DESTDIR=/builddir/build/BUILDROOT/ruby-mysql-2.8.2-9.fc20.i386
 make: *** No rule to make target `/builddir/build/BUILDROOT/ruby-mysql-2.8.2-9.fc20.i386/usr/include/ruby.h', needed by `mysql.o'.  Stop. 
 
Apparently, the mkmf creates Makefile, where everything depends on DESTDIR (see the diff of makefiles here [1]). This is wrong IMO, since obviously Ruby's header files location has nothing to do with location I'd like to install my libraries.

Could you fix this issue? Or if that was always wrong approach, could you provide us better one?

Please note that this was originally reported here: https://bugzilla.redhat.com/show_bug.cgi?id=921650

[1] https://gist.github.com/voxik/5186924
=end



-- 
http://bugs.ruby-lang.org/