[gdal-dev] Backward incompatible changes in the OGR style string format (r19724)

[Daniel Morissette]

Tamas Szekeres wrote:
> 
> I don't think we extensively use filenames and URLs in the style strings
> however we may indeed consider the dot '.' as a reserved char for
> example in the parameters which could be a float value in the style
> strings. In such sense the comma ',' could also be a character which
> already used as a separator in the style strings. By all means if we go
> ahead with this variant, the upper level libraries using the style
> strings (like mapserver) should also be fixed. We may consider whether
> such fix should be appied only in the development version only or in the
> stable branch as well.
>  

You've got a good point that in addition to trunk, at a minimum the the
MapServer 5.6.x branch should be updated with the fix as well... it
would be a little on-liner #ifdef with little/no chances of breaking
anything so there is no risk in doing this in a stable branch.


> 
> BTW: How is the MITAB upstream project normally updated with the changes
> in the GDAL project. Is that based on the SVN changeset in GDAL (in the
> mitab subdir) or based on the bugs submitted to the maptools tracker?  I
> can see a variety of changes/fixes even in the GDAL stable branch which
> should also be applied in MITAB before pulling the whole mainstream back
> in the GDAL SVN.
> 

Updates have to be committed upstream, ideally by submitting a ticket at
http://bugzilla.maptools.org/ ... eventually I should open the MITAB CVS
repository to more GDAL committers to facilitate this process.

I have to admit that as new GDAL committers have started making changes
to ogr/ogrsf_fmrts/mitab without making sure the fix is propagated
upstream we've kind of lost control and are in front of a bit of a mess
at the moment.

Daniel
-- 
Daniel Morissette
http://www.mapgears.com/