can not get tileindex working

[Xin]

Hi Steve,

Thank you so much for your reply! It was very informative and easy to 
understand.

I think I shall stick with idx-ing them for now. I've got a layer for roads 
which is 100megs! I'll attempt tileindexing of the second kind on this later 
if need be.

Thank you so much again.
Xin

On 05/09/05, Stephen Woodbridge <woodbri at swoodbridge.com> wrote:
> 
> Hi Xin,
> 
> It is a little more complicated than you have tried, I will try to
> explain it.
> 
> First, don't worry about shp2tile, most people do not need it and you
> would need to get the basic mapserver working and the tileindexes
> working first, before you can assess if you need it.
> 
> For performance you need to make sure all *.shp files have a *.qix file
> find /path/to/data -name "*.shp" -exec shptree {} \;
> These are spatial indexes and not having these on all files is the
> primary reason for poor performance. Do this first and see how you
> performance is.
> 
> Tileindexes are a way of group lots of similar files together into a
> single layer. This is most useful if you have thousands of shapefiles
> that represent the roads (or whatever) of a country and each file is in
> a separate directory say organized by province or region. Then you can
> create one layer and use the tileindex to collect all the files together
> and control them in a single layer.
> 
> Tileindexes have constraints on them as follows:
> 1) they should generally be of the same shape type (as you mentioned)
> 2) all the files referenced in an index MUST have the same attribute
> structure. This means that if the first file added to the index has 4
> attribute columns say, for example, (int,string,string,double) with
> names (GID,NAME,TYPE,AREA), then ALL files added to the index must have
> the SAME columns in the SAME order.
> 
> Tileindexes are also used in some cases where you have done all the
> above, but you have a really HUGE shapefile. HUGE being lots of objects,
> like 100s of thousands, that a spatially diverse. Then you can use
> shp2tile to chop the large file into lots of smaller files where the
> objects are grouped spatially. Then you use a tileindex to bring them
> all together again into a single layer. If you do this don't forget to
> create *.qix files for all the new shapefiles. I also recommend that you
> put all the tiled files into a directory, like roads.shp into roads/
> 
> Hope this helps,
> -Steve W.
> http://imaptools.com
> 
> Xin wrote:
> > Hi list,
> >
> > I've searched the archives endlessly, and though there's been alot of
> > posts on this I still can't seem to get the hang of it.
> >
> > I'm trying to index my shapefiles, as I've been told this will give it a
> > performance boost. All my data are all in one directory, and they're
> > roughly half a gig in size. It is operating rather slowly at the
> > moment, taking a few seconds to display all the layers.
> >
> > This is how I initially tried to use tile index on my data:
> > ogrtindex *.shp tileindex.shp # to index all my shapefiles
> > in the map file, add below to every layer:
> > TILEINDEX "tileindex.shp,0" # and remove DATA tag
> >
> > This doesn't work!
> >
> > I've since read you have to group different types together into one
> > index file. i.e. all the POINTS in one, LINES in another etc. So you
> > would have say 4 tile index files, each corresponding to each type.
> > Then would I use these filenames, i.e. TILEINDEX "pointindex.shp". Is
> > this the right way, anyway to automated? And what are shptree used for?
> >
> > I tried to use imaptools.com <http://imaptools.com> <
> http://imaptools.com>'s shp2tile, but the
> > library it depends on(shapelib) wouldn't compile.
> >
> > One last question. What's qix and should I use it?
> >
> > Thank you inadvance for your help.
> >
> > Regards,
> > Xin
> >
> 
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-users/attachments/20050905/6f59370b/attachment.html