[mapserver-dev] mapmygis.c

[Steve Lime]

Hi Atilla: I thought most or all of the MySQL join code was in mapjoin.c along
with the DBF and CSV drivers. I would agree that functionality is relevant. Do 
you know what bits of mapmygis.c would need to be preserved?

Steve

>>> Attila Csipa <plists at prometheus.org.yu> 11/05/08 5:51 PM >>>
On Tuesday 04 November 2008 20:50:47 Paul Ramsey wrote:
> - It has no maintainer.
> - It operates on a model of spatial in MySQL (features in blobs) that
> has been super-ceded by the official spatial implementation that has
> now been there for 5 years.
> - It confuses people into thinking that *it* is the real MySQL spatial
> support in Mapserver, when the real support is via OGR.
> - It is deeply ugly code, and now that the mappostgis.c code it was
> based on is retired, I can freely say that. The longer it is there,
> the more likely some unwitting soul will template something on it.
> (See mapmssql2008.c)

I happen to be the original author, and agree on most points, the spatial 
aspect of mygis has been obsoleted years ago and it sure does show plain old 
C not being in my top 5 languages. 

The (perhaps) only important aspect of the whole module that was not mentioned 
here is the ability to do joins to non-spatial mysql columns, even though it 
might not be obvious that it's mygis that actually enables this feature. I 
know OGR provides similar functionality (at least to postgre databases), I'm 
not sure if mysql joins are supported, someone with more ogr+mysql experience 
might chime in on this. If that works, I guess the mygis code can be removed 
with close-to-zero functionality loss. As mygis was never a 'by default' 
build option I would be surprised to see it in use in a significant number of 
installs.
_______________________________________________
mapserver-dev mailing list
mapserver-dev at lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapserver-dev