Alastair Reid wrote:

Is there some way we can tweak things so that Amy's problem (attached) is less likely to happen?

What I'm thinking is that, though stock Darwin systems apparently don't have ncurses, they probably have something equivalent. Could we figure out what that something is, make whatever changes Hugs needs (hopefully none) so that it can use that something and then (re)order the tests in configure.in so that it tests for the preferred form first.

Or, since the problem presumably comes from having built the binary distro on quite a well configured development machine, does anyone out there have access to a less well configured Darwin box that we could use for building binary distributions?

[Oh, how I long for the days when source distributions were the standard way of doing things. You ran configure, it detected what your machine was actually like, and you built a binary which actually matched your machine. Easy! The binary distribution approach only seems to work well when machines are relatively uniform.]

-- Alastair

Well, since MacOS X 10.2 (Jaguar) comes with ncurses preinstalled, the problem is actually due to an old OS version. Even MacOS X versions prior to 10.2 contain ncurses, although included in libSystem (Library Reorganization Considered Harmful!). All Hugs binaries for MacOS X have used a statically linked version of libreadline to avoid external dependencies. So it's basically a question of what what OS versions should be supported. Compiling for 10.1 requires 10.1 installed, and I don't have that anymore. Maybe someone still running 10.1 or 10.0 will want to provide Hugs binaries for those systems? -- Johan

Amy Barnes writes:

...

Finally worked out my problem was not that I didn't know to load hugs, I have a file missing -

[localhost:~] amy% hugs dyld: hugs can't open library: /usr/lib/libncurses.5.dylib (No such file or directory, errno = 2)

---

Hugs-Bugs mailing list Hugs-Bugs@haskell.org http://www.haskell.org/mailman/listinfo/hugs-bugs