The biggest problem with the potentially extremely attractive and valuable Communicating Application Specification announced up in Hillsboro, Oregon last week by Intel Corp and Digital Communications Associates Inc, Alpharetta, Georgia (CI No 996), is that the partners haven’t released full details of the specification yet – although Intel is inviting people to contact its Personal […]
The biggest problem with the potentially extremely attractive and valuable Communicating Application Specification announced up in Hillsboro, Oregon last week by Intel Corp and Digital Communications Associates Inc, Alpharetta, Georgia (CI No 996), is that the partners haven’t released full details of the specification yet – although Intel is inviting people to contact its Personal Computer Enhancement Operation in Hillsboro if they want want copies. The Communicating Applications Specification is described as a set of guidelines to enable MS-DOS – and later, OS/2 – micros to communicate in background with very little user intervention. It includes an application program interface for developers to write to, facilitating their task as much as that of the user. Digital Communications emphasises that it is not a product, hardware, software or service, but an open description that the partners plan to distribute throughout the industry and hope will become a standard on a par with the Lotus Intel-Microsoft memory expansion specification. The aim is to make invoking communications as simple as invoking a printer, so that, once the communications command has been given, data will be transferred in background in the same way that a printer looks after itself once it’s asked to print a file, and the user carries on with other work. The definition includes the core functions of communications scheduling and execution, so that developers can implement it just as they implement a print command in an application. It sits atop the existing communications protocols, effectively automating the procedures that the user presently has to go through manually, and relegating communications to background, enabling them to be invoked from within a word processor or whatever. The developer initiates the communications request by activating a Resident Scheduler and indicating a target or distribution list, the documents to be transmitted and desired time of communication. The Scheduler can access all the communications technologies within the machine without any further intervention, so that the developer – and the user – need know nothing about these. The Scheduler gets a call from the application and finds the necessary information in an electronic phone book, software for maintaining which will be supplied by Intel. The directory will indicate the appropriate communication method and the route to each destination – typically a number for a fax machine in the first version. Transfer Agent A Transfer Agent, comprising code to set up a call, transmit, and close down a call, then transmits the message. The Transfer Agent definition includes enough error checking and correcting code to ensure that the data is sent successfully, and reports back to the sender. The Transfer Agent is loaded by the application onto the 80188 co-processor on the Intel Connection CoProcessor board, so that the work can carry on in background without disturbing the main processor further. Version 1.0 of the specification was released last week and is concerned with high-speed – 9,600 bps – asynchronous modem communications for file transfer, and facsimile transmission. Version 2.0, which will be released in early 1989, will deal support for additional types of hardware and Transfer Agents to other types of communications services such as synchronous communications with mainframe electronic mail services, and remote services such as MCI Mail. Using the initial Intel board, the thing can be used only for communicating with a list of Group III facsimile machines and with a list of other personal computers with the board fitted. Although it is currently pitched at MS-DOS users, Intel stresses that OS/2 users will need it even more, since running a 2,400 bps send in background using OS/2 slows everything else down by an alleged 25%, so that if you tried to send at 9,600 bps, you might as well forget your other applications for the duration. That is as much as the partners are saying at this stage; the specification will be in the public domain, though the pair will le
gally own it.