[mapserver-dev] MapServer 5.2 Beta 3 Available

Paul Ramsey pramsey at cleverelephant.ca
Fri Jun 27 17:46:59 EDT 2008


OK, between the two patches, then, we're in good shape (assuming my
patch worked, jeff?). Yes, I addressed your larger shapetile patch and
totally missed the mapraster.c issue, rah! Your patch will fix. Still
leaves a behavior change, but I'm prepared to believe it's not
something anyone will run across.

P.

On Fri, Jun 27, 2008 at 2:33 PM, David Fuhry <dfuhry at acm.org> wrote:
> Paul,
>
>   I'm under the impression the discussion is about raster tileindexes, not
> shapefile tileindexes.  I'm more ignorant about the latter, and its patch
> was riskier and more invasive, so I have no problem with whatever
> compromises have to be made with the shapefile tileindexes patch (including
> not merging for 5.2).
>
>   With raster tileindexes however, I think the changes are realistic; I
> attached a two-line patch to http://trac.osgeo.org/mapserver/ticket/2369
> (also attached here).
>
>    Jim, does the attached patch fix the Image handling error for you?
>
> Thanks,
>
> Dave
>
>
> Paul Ramsey wrote:
>>
>> David, I agree, it's the relative shapepath case that is failing.
>>
>> You may not like this but I replicated the old logic and just added
>> yours to the top. I think that your idea that in general things are
>> either absolute paths or work is correct, so this ugly code should not
>> be too slow. And I like that it actually falls back: if the file is
>> missing in one place, it'll check another.
>>
>> Patch attached.
>>
>> On Fri, Jun 27, 2008 at 1:53 PM, David Fuhry <dfuhry at acm.org> wrote:
>>>
>>> Paul,
>>>
>>>  For an absolute shapepath, the shapepath clobbers tiFileAbsDir (the
>>> absolute path of the tileindex file's directory), and I think things work
>>> correctly.
>>>
>>>  But the relative shapepath case is not handled correctly.  I'm thinking
>>> that the correct logic needs a branch, and should look like this:
>>>
>>>   if(layer->tileindex) {
>>>     if (map->shapepath)
>>>     {
>>>       msBuildPath(szPath, map->shapepath, filename);
>>>     } else {
>>>       msBuildPath(tiAbsFilePath, map->mappath, layer->tileindex); /*
>>> absolute path to tileindex file */
>>>       tiAbsDirPath = msGetPath(tiAbsFilePath); /* tileindex file's
>>> directory */
>>>       msBuildPath(szPath, tiAbsDirPath, filename);
>>>       free(tiAbsDirPath);
>>>     }
>>>   } else {
>>>     msBuildPath3(szPath, map->mappath, map->shapepath, filename);
>>>   }
>>>
>>>
>>>  Is the lower else clause still ok?
>>>
>>> Thanks,
>>>
>>> Dave
>>>
>>> Paul Ramsey wrote:
>>>>
>>>> This doesn't seem right to me:
>>>>
>>>>     msBuildPath3(szPath, tiFileAbsDir, layer->map->shapepath, filename);
>>>>
>>>> If shapepath isn't set, you get the right answer, but if it *is* set,
>>>> you get something very odd indeed.
>>>>
>>>> P.
>>>>
>>>> On Fri, Jun 27, 2008 at 12:50 PM, Jeff McKenna
>>>> <jmckenna at gatewaygeomatics.com> wrote:
>>>>>
>>>>> Jim,
>>>>>
>>>>> There was indeed a change in beta3 regarding TILEINDEX layers, but it
>>>>> should
>>>>> not affect any existing indexes with SHAPEPATH set, which is your case.
>>>>>  I
>>>>> can duplicate your error with a file structure exactly like yours: with
>>>>> beta2 no error, with beta3 i get your exact errors.  Would you mind
>>>>> reporting this in the associated ticket?
>>>>> http://trac.osgeo.org/mapserver/ticket/2369  Also please include what
>>>>> OS
>>>>> you
>>>>> are testing on (I am testing on Windows).
>>>>>
>>>>> Good find!
>>>>>
>>>>>
>>>>>
>>>>> ---
>>>>> Jeff McKenna
>>>>> FOSS4G Consulting and Training Services
>>>>> http://www.gatewaygeomatics.com/
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On 27-Jun-08, at 11:54 AM, Jim Klassen wrote:
>>>>>
>>>>>> All,
>>>>>>
>>>>>> Beta3 seems to have broken my tiled raster layers. These layers all
>>>>>> worked
>>>>>> fine up through 5.2.0-beta2. Is there something I need to change in my
>>>>>> config?
>>>>>>
>>>>>> The error I get is "msDrawMap(): Image handling error. Failed to draw
>>>>>> layer named 'L10'. msDrawRaster(): Unable to access file.
>>>>>> L10/512000_0-1024000_512000.jpg using full path
>>>>>> L10/512000_0-1024000_512000.jpg"
>>>>>>
>>>>>> In map I have SHAPEPATH set to "PWTS/2006_00"
>>>>>> In the layers I have TILEINDEX "L10/TILEINDEX"
>>>>>> In the tileindex I have Locations such as
>>>>>> "L10/512000_0-1024000_512000.jpg"
>>>>>>
>>>>>> The file system layout is:
>>>>>>
>>>>>> ./AERIAL_2006/ramsey_2006.map
>>>>>> ./AERIAL_2006/PWTS/2006_00/L10/TILEINDEX.{shp,shx,dbf}
>>>>>> ./AERIAL_2006/PWTS/2006_00/L10/*.jpg
>>>>>>
>>>>>> Jim Klassen
>>>>>> City of St. Paul
>>>>>>
>>>>>>>>> "Steve Lime" <Steve.Lime at dnr.state.mn.us> 06/26/08 10:00 PM >>>
>>>>>>
>>>>>> - mapshape.c: Applied patch to make the location of tiled data
>>>>>> relative
>>>>>> to
>>>>>> the
>>>>>>  tileindex directory if SHAPEPATH is not set. (#2369)
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> mapserver-dev mailing list
>>>>>> mapserver-dev at lists.osgeo.org
>>>>>> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>>>>>
>>>>> _______________________________________________
>>>>> mapserver-dev mailing list
>>>>> mapserver-dev at lists.osgeo.org
>>>>> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>>>>>
>>>> _______________________________________________
>>>> mapserver-dev mailing list
>>>> mapserver-dev at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>


More information about the mapserver-dev mailing list