I will use the files to store my database DDL on svn and compare
versions with beyond compare.
It is much easier to use BC folder and file compare instead one single
file with many megabytes (complete extract).
So we will compare database versions, database instances (each
customer has an instance of our database).
It is also easier to database procedure developers to post on svn only
the files their are working on using a dump/downloader from their
development database.
But ISQL produces only a very large single file.
http://tracker.firebirdsql.org/browse/CORE-3685

I've tested your compare database utility. It looks awesome, but for
us is a bit of unusable because it doesn't use the CREATE OR ALTER
VIEW command for FB 2.5. And dropping all view dependencies would
create a too huge script file.
Our database make heavy use of editable view and procedures to enforce
business rules and on some tests a change on a single procedure with a
dependent view lead to lots of dropping and recreating.

Today I already use my own procedures to achieve the DDL on SVN feature.
procedure_PROCEDURENAME_header.sql
procedure_PROCEDURENAME_body.sql
trigger_TRIGGERNAME_header.sql
trigger_TRIGGERNAME_body.sql
view_VIEWNAME_body.sql

Take a look. The field PROCEDURE_EMPTY is saved in a file
XXX_header.sql and the field PROCEDURE_FULL is saved in a file
XXX_body.

create or alter procedure DYPEDEV_PROCEDURE_DOWNLOAD (
 PROCEDURE_TO_DOWNLOAD varchar(255))
returns (
 PROCEDURE_NAME varchar(255),
 PROCEDURE_EMPTY blob sub_type 1 segment size 80,
 PROCEDURE_FULL blob sub_type 1 segment size 80)
AS
declare PARAM varchar(255);
declare PARAM_NO integer;
declare PARAM_QTD integer;
declare PARAM_TYPE varchar(255);
declare PROCEDURE_TYPE smallint;
begin
 PROCEDURE_NAME = trim(PROCEDURE_TO_DOWNLOAD);
 PARAM_QTD = coalesce((select count(*) from RDB$PROCEDURE_PARAMETERS
PP where PP.RDB$PROCEDURE_NAME = :PROCEDURE_NAME and
PP.RDB$PARAMETER_TYPE = 0), 0);
 PROCEDURE_EMPTY = 'create or alter procedure ' || PROCEDURE_NAME;
 if (PARAM_QTD > 0) then
 begin
 for select
 PARAMETER_NAME,
 PARAMETER_NUMBER,
 PARAMETER_SOURCE
 from
 DYPEDEV_PROCEDURE_GET_PARAM(:PROCEDURE_NAME, 0)
 order by
 PARAMETER_NUMBER
 into
 :PARAM,
 :PARAM_NO,
 :PARAM_TYPE
 do
 begin
 if (PARAM_NO = 0) then
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ' (
';
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ' ' || PARAM || ' ' || PARAM_TYPE;
 if (PARAM_NO <> PARAM_QTD - 1) then
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ',
';
 if (PARAM_NO = PARAM_QTD - 1) then
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ')';
 end
 end
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || '
';
 PARAM_QTD = coalesce((select count(*) from RDB$PROCEDURE_PARAMETERS
PP where PP.RDB$PROCEDURE_NAME = :PROCEDURE_NAME and
PP.RDB$PARAMETER_TYPE = 1), 0);
 if (PARAM_QTD > 0) then
 begin
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || 'returns (
';
 for select
 PARAMETER_NAME,
 PARAMETER_NUMBER,
 PARAMETER_SOURCE
 from
 DYPEDEV_PROCEDURE_GET_PARAM(:PROCEDURE_NAME, 1)
 order by
 PARAMETER_NUMBER
 into
 :PARAM,
 :PARAM_NO,
 :PARAM_TYPE
 do
 begin
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ' ' || PARAM || ' ' || PARAM_TYPE;
 if (PARAM_NO <> PARAM_QTD - 1) then
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ',
';
 if (PARAM_NO = PARAM_QTD - 1) then
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || ')
';
 end
 end

 PROCEDURE_EMPTY = PROCEDURE_EMPTY || 'as
';

 select
 P.RDB$PROCEDURE_TYPE,
 P.RDB$PROCEDURE_SOURCE
 from
 RDB$PROCEDURES P
 where
 P.RDB$PROCEDURE_NAME = :PROCEDURE_NAME
 into
 :PROCEDURE_TYPE,
 :PROCEDURE_FULL;

 PROCEDURE_FULL = PROCEDURE_EMPTY || PROCEDURE_FULL || ';';
 PROCEDURE_EMPTY = PROCEDURE_EMPTY || 'begin
' || decode(PROCEDURE_TYPE, 2, '', 'suspend;
') || 'end;';

 suspend;
end;