Changes between Initial Version and Version 1 of Ticket #3975, comment 2


Ignore:
Timestamp:
Jan 11, 2018, 6:50:55 AM (7 years ago)
Author:
pramsey

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #3975, comment 2

    initial v1  
    11OK, I can see two fixes right now, both equally ugly in their own way.
    22
    3 Fix #1:
     3Fix 1:
    44
    55* Change `st_transform` and `postgis_transform_geometry` into SQL wrappers move the actual C bindings into `_st_transform` and `_postgis_transform_geometry`.
    … …  
    77* Potentially cleaner but still ugly variant: change the signature of `st_transform` and `postgis_transform_geometry` to have an extra 'schema' parameter that has a default value of @extschema@, then we can avoid the extra level of indirection and get the install schema injected still. Depends on defaults working and involves changing signature of existing heavily used function.
    88
    9 Fix #2:
     9Fix 2:
    1010
    1111* Go into the C guts of `st_transform` and `postgis_transform_geometry`.
    … …  
    1313* In the PgSQL syscache, look up the namespace that is associated with that function Oid.
    1414
    15 Both fix #1 and #2 will require changing a lot of proj lookup code to take not just the SRID number being looked up, but also the schema to look for spatial_ref_sys in.
     15Both fix 1 and fix 2 will require changing a lot of proj lookup code to take not just the SRID number being looked up, but also the schema to look for spatial_ref_sys in.