[Linux-ha-dev] SWIG?

[Horms]

On Wed, Jan 31, 2007 at 12:10:58PM +0000, David Lee wrote:
> On Wed, 31 Jan 2007, Lars Marowsky-Bree wrote:
> 
> > On 2007-01-31T10:40:34, David Lee <t.d.lee at durham.ac.uk> wrote:
> >
> > > These days, all that is commented out, so configure runs to completion,
> > > then 'make' falls in a heap in "lib/mgmt" when it can't find swig.  Ouch!
> >
> > You might have success if you build without the GUI; it is built by
> > default, but requires SWIG. If you disable it, it should work.
> 
> Thanks, Lars.
> 
> I could try that.  But my perspective isn't "getting it to work just for
> me".  Rather it is "ensuring it will work for all the end-users out there,
> of which I happen to be one example".
> 
> What I've come across is a technical inconsistency that needs to be
> resolved within the code, for the sake of potentially many end-users out
> there.
> 
> As an overall philosophy:
> 
> 1. If a compulsory OS-like component is absent, 'configure' should detect
> this and fail (with informative error message).
> 
> 2. If 'configure' successfully completes (including if OS-like components
> that are desirable but not strictly necessary are lacking) then 'make'
> MUST aim to complete successfully (any post-configure errors are bugs).
> 
> 
> At present, we failing to handle the absence of swig consistently.  All I
> want to do (for the sake of others) is to resolve that inconsistency
> cleanly within the code.
> 
> For this particular bug, the fundamental question is "is SWIG now
> absolutely compulsory?"  If so, then configure should scream to a halt if
> swig is absent.  If not, then "lib/mgmt" needs to be able to do something
> useful without swig (or should we even be building "lib/mgmt" in the first
> place?).

I suspect that this is more of an oversight that anything else.

-- 
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/