One question... >From a performance perspective, I would have thought that having the fliter FILTER (stype='B' and uc='%vucv%') inside the DATA statement SQL would be faster, as only the relevant records would be returned to mapserver, & Postgres would be more efficient at doing this.
Is there any advantage in using FILTER instead of an SQL WHERE clause for this? Brent Wood --- On Tue, 8/30/11, Julien Cigar <[email protected]> wrote: From: Julien Cigar <[email protected]> Subject: Re: [mapserver-users] Issue with run-time substitution? To: [email protected] Date: Tuesday, August 30, 2011, 11:49 PM On 08/30/2011 13:43, Stephen Davies wrote: > I am running mapserver 6.0.1 with map files migrated from 5.6.3. > > One of my map files has stopped working properly and I suspect that it may be > a > bug in the 6.0.1 run-time substitution code. > > The relevant map file layer sections look like this: > > LAYER > CONNECTIONTYPE postgis > NAME "battery" > DATA "geom from wmd using unique id using SRID=4283" > CONNECTION "user=scldad dbname=benparts" > STATUS ON > TYPE POINT > VALIDATION > vucv '[0-9]*_[0-9]*' > END > FILTER (stype='B' and uc='%vucv%') > PROJECTION > "init=epsg:4283" > END > MAXSCALE 5000000 > LABELITEM "label" > CLASSITEM "state" > CLASS > EXPRESSION "G" > STYLE > COLOR 0 255 0 > SYMBOL 'dot' > SIZE 7 > OFFSET 17 0 > END > LABEL > POSITION CR > TYPE TRUETYPE > FONT arial > SIZE 8 > COLOR 0 255 0 > OFFSET 20 0 > FORCE TRUE > STYLE > GEOMTRANSFORM 'labelpoly' > COLOR 255 255 255 > END > END > END > . > . > END > > There are twelve layers with essentially the same definition apart from the > STYPE values. > > The invoking URL has&vucv=137_11 but the postgresql log reveals that no > substitution has occurred. The final pgsql command still has '%vucv%'. > > I suspect that having more than one FILTER with the same substitution breaks > things as all other substitutions work fine. > > (I also tried putting the substitution in the DATA entry but with no better > result.) You should use a METADATA section, something like: METADATA "vucv_validation_pattern" '[0-9]*_[0-9]*' END instead of VALIDATION > > Cheers, > Stephen > -- No trees were killed in the creation of this message. However, many electrons were terribly inconvenienced. -----Inline Attachment Follows----- _______________________________________________ mapserver-users mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/mapserver-users
_______________________________________________ mapserver-users mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/mapserver-users
