On Feb 13, 2008 6:15 PM, Stefan Gofferje <<a href="mailto:gradsusers@gofferje.homelinux.org">gradsusers@gofferje.homelinux.org</a>> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Arlindo da Silva schrieb:<br><div class="Ih2E3d"> > Important Note: The sources for grads-2.0.a0 released by COLA *do not*,<br>> repeat *do not*, build out of the box with the supplibs-2.0.1, although<br>> it contains all the necessary libraries.<br>
<br></div>It could help to do something like this:<br><br>for a in */*.h ; do b=$(echo $a | cut -d"/" -f2); ln -s $a $b ;done<br><br>in the include directory of the supplib-src dist for building grads2. :)<br></blockquote>
<div><br>A word of caution: be *very careful* doing this sort of thing, unless you know exactly what you are doing. You can do this flatenning of the header files for most packages, except for packages nc-dap, netcdf, netcdf4 and in some cases hdf-4 (when netcdf is not disabled): all these packages implement the netcdf API and have header files of the same name. Not only you have to keep them apart, but you also have to be careful not to include "netcdf.h" from the netcdf package when building gradsdods, for example. The problem of mixing and matching these header files is that the bugs can be very subtle: everything builds and work 90% of time, and it can be extremely hard to debug. For this reason, I reorganized the supplibs to keep all the header files in separate directories for each package. It also makes it easier to upgrade individual packages.<br>
<br> Arlindo<br><br></div></div>-- <br>Arlindo da Silva<br><a href="mailto:dasilva@alum.mit.edu">dasilva@alum.mit.edu</a>