On Fri, 22 Apr 2005, David Berry wrote:
> > these failures (and skipping of CONVERT and TSP) seem to be caused by
> > recent changes to ast_link. Is it the case that we should now be using
> > `ast_link -fsla` to retain the old behaviour?
>
> Looking at figaro, it looks like the problem may well be caused by the
> changes to ast_link, but I would say the problem lies in figaro rather
> than ast. What I did yesterday was to change the ast_link script so that
> it no longer pulls in the fortran slalib library by default. This is
> because it now uses its own GPLed slalib code.
>
> Figaro code makes direct calls to fortran slalib and so should have
> `sla_link_adam` in its link script, but it doesn't. It has previously been
> relying on `ast_link_adam` dragging in slalib, which is a bit naughty.
> Not sure if the same goes for the other components which are failing.
Actually GAIA is failing because you've pulled the Fortran libraries out
from under it. CONVERT and TSP because they depend on FIGARO.
> So, true this could be fixed by using `ast_link_adam -fsla`, but I
> would say the proper fix for this is to add `sla_link_adam` to the
> figaro_link_adam script, etc.
>
> Any disensions?
So does using `sla_link_adam` also require `ast_link -fsla` to avoid
clashing symbols? Or will you only see that when linking against the C
version of SLALIB?
Peter.
|