>Dear all,
>I'm having a problem with a backend process with terminal services clients.
>I've written a backend standalone exe (olepublic) that runs upon startup of the server.
>Clients login via terminal services to do their operation and as part of the operation, the individual sessions will send in processing jobs to a table (as a new record) and the processing will then be done by the backend process. A temporary table containing the result will be created with a filename that is given by the client process. On top of this, records are updated to databases in other directories.
>The problem that I have is that this model works fine from local network clients. However, for clients accessing via terminal services, each session needs to startup the backend standalone exe (even though one has already been started under the administrator logon) in order for the client to have access to the backend process.
>Is there a solution to this ?

Can you be more specific? What is actually the problem, what is it that the client process cannot do?

As I understand this, the client doesn't need any access to the backend process - it needs only to be able to write a new record in the jobs table, and to check when is the job done (by either checking the status of the job record, or by checking for the existence of the result table). Is it that the client doesn't see the result table because it's owned by the administrator (process which creates it is logged in as admin)? If that's the case, the whole table can be stuffed into a memo of the jobs table, and then the client would be able to strtofile() it out - easier than to fiddle with the security attributes of the result table.