[mapserver-users] testing mapserver large files rendering improvements

[Bart van den Eijnden (OSGIS)]

Want to toss in my 2 cents as well.

I really find it a PITA what can happen if your qix file is out of sync 
with your shapefile. Very unpredictable things can happen, so I even 
find this a bigger problem than no qix file at all to be honest. A 
msDebug for this would be very much welcome.

TatukGIS does the same when it opens up a shapefile, it checks if the 
spatial index is up to date and existing, if not it will update/create 
it. I think this makes sense, and a compile option would be a reasonable 
way to control this. A few questions though:

1) how will Mapserver know by default which depth to use for the index? 
On large point shapefiles the default depth can result in negative 
performance.
2) currently shp2img does not report any error when it does not have 
write permission on Linux, but I guess the new approach will first check 
write permission before doing any analytics?

So it might make sense to have a layer setting to overrule the default 
depth? Or to overrule automatic creation at all for a specific layer?

Best regards,
Bart

Stephen Woodbridge wrote:
> Paul,
>
> I think this is the best option because it gives control to the 
> package builders and anyone that is a power user.
>
> I would also like to see msdebug report missing qix files. This is a 
> valuable tuning tools for power users.
>
> Thanks,
>   -Steve
>
> Paul Ramsey wrote:
>> I have another "third way" option, which would be a compile-time
>> directive (--with-auto-qix) so that packagers can make dummy-friendly
>> builds if they so desire.
>>
>> ??
>>
>> P
>>
>> On Thu, May 15, 2008 at 9:25 AM, Jeff McKenna <jeffmckenna at gmail.com> 
>> wrote:
>>> My vote is for Paul's default qix creation option, by MapServer.  
>>> Let's make
>>> life easy for those beginner Windows users.
>>>
>>> -jeff
>>>
>>>
>>>
>>>
>>>
>>> On 15-May-08, at 12:07 PM, Frank Warmerdam wrote:
>>>
>>>> Daniel Morissette wrote:
>>>>> Guillaume Sueur wrote:
>>>>>>> In my mind, we should do the right thing by default, and give the
>>>>>>> power users the option to opt out, rather than forcing everyone to
>>>>>>> become a power user to achieve something as basic as indexed file
>>>>>>> access.
>>>>>> +1 for this approach
>>>>>>
>>>>> The right thing to do is not the same in everyone's eyes.
>>>> Folks,
>>>>
>>>> I'm also against mapserver automatically creating .qix files, for the
>>>> reasons
>>>> Daniel and others have given.
>>>>
>>>> I'd add that current shapefile update tools (including he mapscript
>>>> shapefile
>>>> writing code, Shapelib and OGR) do *not* update .qix files.  If 
>>>> secretly
>>>> create a .qix file for a shapefile that is being periodically 
>>>> updated all
>>>> hell is likely to break out.
>>>>
>>>> BTW, .qix files can be *relatively* large.  For a shapefile of 
>>>> points, the
>>>> .qix can easily be bigger than the .shx file.
>>>>
>>>> Best regards,
>>>> -- 
>>>>
>>> _______________________________________________
>>> mapserver-users mailing list
>>> mapserver-users at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>>>
>> _______________________________________________
>> mapserver-users mailing list
>> mapserver-users at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>
> _______________________________________________
> mapserver-users mailing list
> mapserver-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>
>


-- 
Bart van den Eijnden
OSGIS, Open Source GIS
bartvde at osgis.nl
http://www.osgis.nl