On Thu, Mar 6, 2014 at 2:53 PM, Antti Kantee <[email protected]> wrote:
> On 06/03/14 14:06, Justin Cormack wrote:
>>> Now, questions:
>>> should this be the default mode of operation?  In other words, should we
>>> use this for the automated tests and docs?
>>
>> Sounds tidier.
>>
>> i don't think I would be inclined to use it for the tests, but
>> probably would for the docs. Test machines might not have bash
>> installed, and then the tests document what is going on, while the
>> docs are more readable.
>
> Probably the only thing that will work differently on a non-bash bourne
> shell is that prompt will be 'rumprun ($RUMP_SERVER)$ ', and we won't
> see that in the tests anyway.  I can test it later when I find a
> non-bash sh from somewhere).

ok.

Actually ash works fine and does replace the vars (thats the default
shell in Debian).

> IMO we should test what we tell the users is going to work.  Also, we
> can run the tests with "set -x" if you want a log of what's going on --
> but I do like the current quietness.
>
>>> What should the relationship of rumprun and rumpremote be?  It seems to
>>> be like they're completely different animals.  Should there be two
>>> separate builds?  For one, rumpremote doesn't require building the
>>> kernel components.  Also, the above type wrapper for rumprun is only
>>> available e.g. via the Lua console, and cannot be implemented using a shell.
>>
>> If you can make a simpler build that is quicker and only builds
>> rumpremote then we could make that the default and use for all the
>> tests. I have some use cases for rumprun, but can always fork it...
>
> Well, hmm, the tests do need the kernel side.  So I guess at least the
> tests would need to build the kernel portions -- I don't think we can
> quite go binary just yet.
>
> I'd also like to see the package eventually becoming installable.  Which
> means cleaning up the mess I've made in the Makefile (should never write
> Makefiles with the assumption that someone who like build systems is
> going to fix them later).  But what about we pencil down that we're
> going to install everything into $PREFIX/lib/rumprun, with the hierarchy
> being:
>
> rumprun/bin/
>    rumpremote wrappers
> rumprun/lib/
>    .so's PLUS librumpclient
> rumprun/internal/
>    rumpremote, rumprun, rumprun.lua
>
> So rumprun would still work, just need to set LD_LIBRARY_PATH (and
> LD_DYNAMIC_WEAK).  The wrappers would just work after you source the env
> file (which is why we need a known location for librumpclient).
>
> Seem ok?

yes. Thats everything except rumpremote.bash which not quite sure
where you would install - its not an executable...

>    - antti
>
> *) I do have two changes I want to make, but given it's been several
> years, not sure if those are going to happen soon

To what?

------------------------------------------------------------------------------
Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.
With Perforce, you get hassle-free workflows. Merge that actually works. 
Faster operations. Version large binaries.  Built-in WAN optimization and the
freedom to use Git, Perforce or both. Make the move to Perforce.
http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
_______________________________________________
rumpkernel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rumpkernel-users

Reply via email to