How to use sde.st_transform (Oracle)?

Question

According to this article, http://resources.arcgis.com/en/help/main/10.1/index.html#/ST_Transform/006z0000009s000000/, to apply file-based CRS transformation in using sde.st_transform function, need to copy the ArcGIS pedata folder to the database server and setup environmental variable PEDATAHOME on the database server. It gives an example for databases on Windows. I am wandering if anybody has done this for SDE database on Linux and what else needs to be done. Does the database instance needs to be rebounced or the listener needs some extra configuration? The st_geometry functions work fine on our system, it is just how we can use file-based CRS transformation, such as from NAD27 to WGS_1984_UTM_Zone_15N with NAD_1927_To_WGS_1984_79_CONUS datum transformation in sde.st_transform function.

AllenGuan1 · Answer

You will certainly need to restart the listener; the need for a database bounce
 is less likely. Unfortunately, you can't prototype all possible problems as thought
 experiments -- At some point you need to try it in your dev server, and see how
 it works out.

- V

Thanks Vince, it seems to me the st_transform function in the cited URL is for ArcSDE 10.1 which allows the the projection to be on a different geographic datum (sde.st_transform (g1 sde.st_geometry, srid integer, geogtran_id integer),  The doc on 10.0 ArcSDE (http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//006z0000009s000000) requires the destination project to be in the same geographic datum as the original one. Our ArcSDE is still in 10.0 so that makes the st_transform not useful in our case. I will test it when moving to 10.1. I like 10.1 much better as it takes EPSG code as SRID.

VinceAngelo · Answer

You will certainly need to restart the listener; the need for a database bounce
 is less likely. Unfortunately, you can't prototype all possible problems as thought
 experiments -- At some point you need to try it in your dev server, and see how
 it works out.

- V

AllenGuan1 · Answer

The process should be the same, but setting environment variables can't be done
at a server level in Unix, and some of the newer Oracle releases seem to clear the 
environment to avoid security pitfalls, so you need to make sure PEDATAHOME is
 being set correctly (before the service start request), and that it's still available 
when necessary (might need to be set explicitly in the listener files).

- V

That is exactly my point. Windows and Linux server variables are not set the same way and I want to make sure whether PEDATAHOME needs to be set before the DBA starts the database instance or if it is possible to set it in the listener file (so only need to rebounce the listener without rebouncing the database). I need to know exactly what needs to be done before asking the Linux DBA to do it. We don't have much room to go back and forth with DBA.