search

reasonml / reason-cli

Globally installable Reason toolchain.
MIT License
290 stars 23 forks source link

MBP install experience #13

Closed glennsl closed 7 years ago

glennsl commented 7 years ago

This is on a first gen retina MBP, which has been upgraded to Sierra but never reinstalled. In other words, it's far from clean.

First attempt resulted in this error:

>> Fatal error: Module CamlinternalLazy unavailable.

Second attempt resolved that simply by being a second attempt, but now it fails with:

*** opam-installer-bin: building from source...
*** substs: building from source...
env: python2: No such file or directory
make: *** [substs.build] Error 127
make: *** Waiting for unfinished jobs....
env: python2: No such file or directory
make: *** [opam-installer-bin.build] Error 127
*** ocaml: building from source...
env: python2: No such file or directory
make: *** [ocaml.build] Error 127

There is no python2 on my system, so I manually symlink it from python27. Still get the same error. Turns out my python is installed with macports which puts its stuff in /opt/local, so I put the symlink beside python27 in /opt/local/bin, and for whatever reason that doesn't seem to jive with the installer. So I uninstall python from macports and reinstall it with homebrew. I've been told that will correctly set up the python2 symlink for me. It doesn't. But i manually symlink it, now in /usr/local/bin and voilá. It works!

But it took a bit more tinkering than I'd like...

jordwalke commented 7 years ago

Thanks, Here's the work to fix this by removing python2 requirement. It's a known issue. Thanks for the patience - reason-cli isn't yet advertised because we haven't got the edge cases like this worked out yet. Soon though.

jordwalke commented 7 years ago

The core issue here is now fixed! No more python required (thanks to @IwanKaramazow ) The "CamlLazy" part is likely going to occur from time to time until we upgrade to 4.04, but it seems very infrequent for me and I install this thing several times a day. Happy to go further and fix that flakiness as well if it is still an issue in practice.