Re: Opam installation woes

[Jonathan Ludlam <Jonathan.Ludlam@xxxxxxxxxxxxx>]

Hmm. Going back to the opam woes, I'm once more having issues - now with the 
standard 'inconsistent assumptions' errors and the like. I suspect this is 
because things are being installed in the wrong places.

The problem I had earlier was that after the switch to the xen compiler, 
~/.opam/system/... was being reported by ocamlc.  Unless I'm misunderstanding 
something, shouldn't the ocamlc binaries be pointing at ~/.opam/xen/... 
instead? It looks to me a lot like stuff was being installed into 
~/.opam/system when it should have been going into ~/.opam/xen - specifically 
the ocamlc binaries ended up in .opam/system/bin. Is that expected behaviour?

Jon



On 7 Aug 2012, at 23:50, Anil Madhavapeddy wrote:

> 
> On 7 Aug 2012, at 23:49, Richard Mortier <Richard.Mortier@xxxxxxxxxxxxxxxx> 
> wrote:
> 
>> 
>> On 7 Aug 2012, at 23:46, Anil Madhavapeddy wrote:
>> 
>>> On 7 Aug 2012, at 23:34, Richard Mortier <Richard.Mortier@xxxxxxxxxxxxxxxx> 
>>> wrote:
>>> 
>>>> (fwiw i've just done this in my fork of opam-repo-dev, and sent a pull 
>>>> req.  jon, by all means give this a go if you haven't fixed it yourself 
>>>> already...)
>>> 
>>> All merged to opam-repo-dev. Btw, I don't seem to get notification of pull 
>>> requests to Mirage organisation repos, so feel free to push directly to 
>>> mirage/opam-repo-dev if it fixes your problem.
>> 
>> ok, will do.
>> 
>>> I think it's high time we setup a little continuous build VM.  Thomas, is 
>>> there an easy way to iterate through every package and attempt a 
>>> compilation? I could do a big 'opam install `opam list`'-style command, but 
>>> something that has reasonable logging for failures would be useful.
>> 
>> 
>> what would you do about package interdependencies in the case of failure?
>> 
>> (sounds like a job for tsort ;)
> 
> there's a SAT solver in OPAM for dependencies. Not my problem ;-)
> 
> -a
>