**Sat 1/30/2010 3:51 PM...PS: At one stage prior to the RES project we had considered using sqlj to interpret the SQL and PLSQL blocks. If all else fails with JDBC it may be an option to consider, however I am not sure that it is available for all DB's supported by Cobol vendors.
Sun 2/7/2010...I am not really certain how many of our programs make use of group items, but I expect some are making use of this for a small proportion of the data elements. The one main verb that would be acting on the group item is the initialise verb to set initial values but after that they would be primarily handled inside the rest of the program as elementary data items.
It might be nice to have some sort of indication from RES how much use of group items there is which might give us an ability to work out if we can swap to using the native java types.
We recently tried out Open Cobol compiler which turns Cobol to C and then compiles C to run on the Unix box. This worked for us up to a point, however if fails to deliver when we tried to get it to work inside an application server. We could not get the DB connection provided by the application server to work from within C code and the Java Native Interface. This meant that each service function would create a connection to the DB and run one or two SQL's and disconnect. ie: It was fine for short tests, but when we make 10,000's of calls a day, this failed to deliver on what we needed. The way it dealt with group items was however quite fast, and although it is hard to tell, it seemed to be quite on par with the native MF Cobol compiler.
The main thing we will find (I think) is that we will want to have the RES converted java programs run within a JEE application server and make use of the connection pooling from within the JEE container. To achieve this, we will have to look at how you have done the connection side of things, and see how easily we can swap the stand alone connection for one provided by the connection pool. We will also write our own code generator so we can expose each of these service functions as a web service. We have an Access DB with the definition of the parameters for each of programs, and a web services wrapper should be simple to generate from the definitions in this database.
**
Anyway, hopefully this gives you more on what we are going to do with RES. Thanks.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Please rate in a scale of 1-5 the following features of RES generated Java code in your project at present.
compilable error free (1-5)
maintainable (1-5)
functionally equivalent(1-5)
Also feel free to add comments/questions/suggestions.
Thanks.
This is from Jim Nicholls, Australia
**Sat 1/30/2010 3:51 PM...PS: At one stage prior to the RES project we had considered using sqlj to interpret the SQL and PLSQL blocks. If all else fails with JDBC it may be an option to consider, however I am not sure that it is available for all DB's supported by Cobol vendors.
Sun 2/7/2010...I am not really certain how many of our programs make use of group items, but I expect some are making use of this for a small proportion of the data elements. The one main verb that would be acting on the group item is the initialise verb to set initial values but after that they would be primarily handled inside the rest of the program as elementary data items.
It might be nice to have some sort of indication from RES how much use of group items there is which might give us an ability to work out if we can swap to using the native java types.
We recently tried out Open Cobol compiler which turns Cobol to C and then compiles C to run on the Unix box. This worked for us up to a point, however if fails to deliver when we tried to get it to work inside an application server. We could not get the DB connection provided by the application server to work from within C code and the Java Native Interface. This meant that each service function would create a connection to the DB and run one or two SQL's and disconnect. ie: It was fine for short tests, but when we make 10,000's of calls a day, this failed to deliver on what we needed. The way it dealt with group items was however quite fast, and although it is hard to tell, it seemed to be quite on par with the native MF Cobol compiler.
The main thing we will find (I think) is that we will want to have the RES converted java programs run within a JEE application server and make use of the connection pooling from within the JEE container. To achieve this, we will have to look at how you have done the connection side of things, and see how easily we can swap the stand alone connection for one provided by the connection pool. We will also write our own code generator so we can expose each of these service functions as a web service. We have an Access DB with the definition of the parameters for each of programs, and a web services wrapper should be simple to generate from the definitions in this database.
**
Anyway, hopefully this gives you more on what we are going to do with RES. Thanks.