First and foremost:
The rpmbuild parser is allergic to "\r" line termination.
You will see some extremely bizzarre problems if you run rpmbuild
on a spec file polluted with \r termination.
The problem has needed fixing for years, just, well, windows and linux
don't really coexist happily because of ideology, and so noone has
On Feb 1, 2008, at 8:22 AM, Peter Kalbus wrote:
> We want to use rpm5.x in an embedded environment. The development
> is a cross-platform environment hosted on win32 systems. To get rpm
> 5.x usable, we need rpm5.x running in that environment.
> On the win32 host, it must be possible to create packages, which
> can be installed on the embedded systems and which can be installed
> on other development systems. This means,that the rmpbuild and rpm
> toolchain, including the database parts must be available on the
> win32 host system.
> In which way rpm5.x is capable be be used on win32 systems?
> Ideally, we're interested in a native port, which does not force
> the installation of any un*x runtime environment.
The majority of the code base of rpm likely ports to Win32 (and any
without too much pain.
Equivalently, where you find major porting pain, the functionality is
likely not critical to rpm
(but perhaps I'm an optimist)
The design of rpm likely has some major problems however. There's
uglix culture buried in cp vs copy, file paths without C: and '//'
UNC conventions, etc
that would need to be addressed in a native port.
Adding items to a registry would need to be added somehow for native
Installing a unix runtime is the typical approach. I can't recall
attempting building rpm natively.
73 de Jeff
Received on Fri Feb 1 16:17:49 2008