[FGS] build_si error?

[Frank Warmerdam]

Daniel Morissette wrote:
> Frank,
> 
> I have the same problem on my system. I stepped through the scripts and 
> found that in my case the problem is a conflict of dependencies. The 
> build script calls "fgsdev list_modules_dep" to create the list of 
> dependencies and if that returns an error then the main script gets 
> confused and the "!" that you see come from the "!" in the error 
> messages. See what I get:
> 
> $ fgsdev list_modules_dep mapserver-php
> ! Conflicting version for 'gdal-base': Another(s) package(s) request(s) 
> 'svn_branch_1.4'
> ! and 'gdal-base' requests 'cvs_HEAD', aborting.
> ! /opt/fgs1/fgs-dev/bin/fgsdev: Fatal error returned, exiting...
> 
> 
> I'd suggest that you try 'gsdev list_modules_dep gdal_python-module 
> gdal_ecw-module mapserver-base libstdc++-lib python_mapscript-module' 
> and fix whatever dependency conflict you may have.
> 
> I'm not sure of the right way to fix the dependency, but in my case I 
> suspect I get this conflict because I switched my GDAL from CVS to SVN 
> but I didn't rerun 'fgsdev build_pkg' for some packages.

Daniel,

Ah, I see.  In fact, I was missing unpacked module directories for
some packages but the effect was similar.  The "!" error messages
were sucked up and evaluated screwing up the dependency checking.

I would like to change list_module_dep.sh and similar scripts to
report errors to stderr (file descriptor 2) like this:

echo "! Module '$MODULE_NAME' doesn't have VERSION file 
'$FGS_MODULES_HOME/$MODULE_NAME/etc/fgs/pkgs/$MODULE_NAME/VERSION'" >&2

That way the error messages are reported and the return code indicates
failure.  Is there any objection to such a change?

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | President OSGeo, http://osgeo.org