My opam package does not seem to install the binary

Hi,

I am trying to release a new OPAM package called electrod (repo). Following the OPAM documentation, I created an electrod.install file (that must install a binary called electrod); but when doing opam pin add electrod ., the package is compiled with success but the electrod program is not copied into the OPAM “bindir”. Any help would be greatly appreciated!

Looking at your repo, the .install file doesn’t appear to be in the right place - are you certain it’s being picked up when you pin?

That was it, thanks. Shame on me!

Easily done!

Incidentally, the intention is that .install files are generated by your build system (so in this case, just a simple echo in the Makefile, obviously)… it’ll make it easier when you upgrade the build system to Dune :wink:

I’d love to switch to Dune (and topkg), but my experiments until now have failed, as I used various packages that didn’t seem to be so easy to deal with (or my understanding of the documentation was not good). (Well it’s also that I don’t program on a daily basis so every new tool takes ages for me to learn.)

Wow so I got this working with Dune and Topkg, but the first version I designed is certainly non-conventional so I would love to get feedback and suggestions: the branch is here.

1 Like

It looks good to me :smile: Was the renaming of the modules in the commit necessary to make it build?

Thank you for the reviewing.
Jbuilder says that Location.cmi and Symbol.cmi can already be found in OCaml’s code, so I solved this name clash this way. All the times I tried Jbuilder in the past, this same problem happened… A problem that I don’t have with Ocamlbuild.

Thank you for the reviewing.
Jbuilder says that Location.cmi and Symbol.cmi can already be found in OCaml’s code, so I solved this name clash this way.

I’m sure I’m being slow reproducing it, but what exactly was the error message? I seem to be able to build with no problem with the much simpler dra27/electrod#80cba6 (which just reverts the renaming of those two files).

Well I can’t reproduce the issue, your fix works :slight_smile:
I will integrate it back then. I can’t recall the exact message but I recall clearly that it complained about Location.cmi being present in compiler-libs, and I already had experienced this a few months back. Anyway, this is fixed! Thank you very much.