This procedure had been written some time ago and it does not take into account that PostgreSQL actually has the "text" datatype. Therefore I propose to patch it (especially as we need that for ]project-open[):
diff --git a/packages/acs-subsite/tcl/attribute-procs.tcl b/packages/acs-subsite/tcl/attribute-procs.tcl
index 6c252df..e33b933 100755
--- a/packages/acs-subsite/tcl/attribute-procs.tcl
+++ b/packages/acs-subsite/tcl/attribute-procs.tcl
@@ -195,7 +195,13 @@ ad_proc -private datatype_to_sql_type {
"number" { set type "number" }
"money" { set type "number (12,2)" }
"date" { set type "date" }
- "text" { set type "varchar(4000)" }
+ "text" {
+ if { "postgresql" eq [db_type]} {
+ set type "text"
+ } else {
+ set type "varchar(4000)"
+ }
+ }
"integer" { set type "integer" }
"enumeration" { set type "varchar(100)" }
"keyword" { set type "varchar(1000)" }
