Re: [Caml-list] Build OCaml on linux

[Francois Berenger <berenger@riken.jp>]

Did someone mentioned ocamlbrew in order to perform installation
automatically?

https://github.com/hcarty/ocamlbrew

It also installs a few useful stuffs (oasis, odb, etc.).
This is what I use in our lab to install OCaml in user's home
in case they have a non Debian-like Linux distro.

Regards,
F.

On 09/25/2012 06:42 AM, Didier Cassirame wrote:
> Hi,
>
> First of, my apologies, I made a few mistakes in my previous message,
> which probably contributed to confuse everyone:
> - the -stdlib parameter doesn't exist, the correct parameter is "-I
> <path/to/stdlib>"
> - I actually cannot tell if the ocamlrun binary relies on a bytecode
> executable version number (that said a version number exists in compiled
> libraries),
>
> Regarding your problem, according to my own tests, it is not safe to
> change the shebang line in bytecode executables. It is
> preferable to explicitely launch a bytecode program with the interpreter
> corresponding to the compiler used to build the program.
>
> Sorry again for the generated confusion,
>
> didier
>
> 2012/9/24 Zhi Han <Zhi.Han@mathworks.com <mailto:Zhi.Han@mathworks.com>>
>
>     Hi,
>
>     Thanks everyone for the help so far. I was going through the
>     Makefile and it appears that there is a native target 'ocamlnat'. If
>     I do 'make ocamlnat' after 'make world.opt' that should create a
>     native executable toplevel that does not require 'ocamlrun', right?
>     It is a little odd because I presume 'make world.opt' would make all
>     native targets.
>
>     The workflow we want to support is:
>     Build OCaml from source on a machine and copy the installation to a
>     server.
>
>     1) Store the OCaml at a shared file server, so that users can setup
>     a few environment variables to use the shared OCaml.
>
>     2) When the users have intermittent network problems, they shall be
>     able to copy the network version to their machine and setup the
>     environment to use the local version.
>
>     Having the byte code really adds complexity to these workflows. Is
>     it safe to simply edit the bytecode to modify the first line?
>
>     Zhi
>
>     -----Original Message-----
>     From: Didier Cassirame [mailto:didier.cassirame@gmail.com
>     <mailto:didier.cassirame@gmail.com>]
>     Sent: Saturday, September 22, 2012 4:59 PM
>     To: Zhi Han; caml-list@inria.fr <mailto:caml-list@inria.fr>
>     Subject: Re: [Caml-list] Build OCaml on linux
>
>     Hello Zhi and Everyone,
>
>     The first line of the bytecode executable is just a hint for the
>     shell to locate the bytecode interpreter. Nothing stops you from
>     running a given bytecode executable with the interpreter of your
>     choice, provided it supports the file revision of said executable.
>
>     So, you have two options :
>     the first one is to keep track of the location of the ocamlrun
>     binary, and prepend every command line running ocaml, ocamlc,
>     ocamlopt with the full path to ocamlrun, i.e. :
>
>     $ /path/to/ocamlrun ocamlc myfile.ml <http://myfile.ml>
>
>     instead of
>
>     $ ocamlc myfile.ml <http://myfile.ml>
>
>     You may have to give the full path to ocamlc as well, and perhaps
>     also indicate the location of the stdlib if they are also moved from
>     their default location (use the -stdlib command line parameter of
>     ocamlc).
>
>     The second option is to recompile the ocaml software from source
>     (you are perhaps already doing that, if you are moving the
>     executables around), specifying the final install location of the
>     binaries at the configuration step (see INSTALL in the source tar ball).
>
>     cheers,
>
>     didier
>
>