SLD against Oracle is broken in 5.0

[Bart van den Eijnden]

Hi list,

SLD support against Oracle is broken in MS 5.0 unfortunately. We've been
debugging the whole morning and have identified the following problems in
the source code (mapogcfilter.c):

1) the classes created by mapogcsld.c are destroyed because of the following
call:

line 919:
initClass(lp->class[0]);

the code already has a comment like which indicates this could be a problem:

    /* TODO: Doesn't this code leak the contents of any pre-existing class
     * in the layer???
     */

2) the if clause in FLTLayerApplyCondSQLFilterToLayer never gets entered
since bOnlySpatialFilter is always true (it is passed in as MS_TRUE always
in the caller function, i.e. return FLTApplyFilterToLayer(psNode, map,
iLayerIndex, MS_TRUE);)

This means Mapserver always evaluates the query, which is very slow for
Oracle, instead of adding a WHERE to the SQL using FILTER. For now we just
changed the if clause:

if (!bOnlySpatialFilter && FLTIsSimpleFilter(psNode))

to:

if (FLTIsSimpleFilter(psNode))

but someone should look at this more carefully I guess ....

3) it's better to always use parantheses around the FILTER, since otherwise
OR and AND won't work correctly as intended, so change line 929 from:

        else //POSTGIS OR ORACLE if (lp->connectiontype == MS_POSTGIS)
          sprintf(pszBuffer, "%s", szExpression);

to

        else //POSTGIS OR ORACLE if (lp->connectiontype == MS_POSTGIS)
          sprintf(pszBuffer, "(%s)", szExpression);

Hope this helps.

Best regards,
Bart
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-dev/attachments/20071205/2e579134/attachment.html