WSPC05429 - Laserfiche WebLink

Home Browse Search

000u37 Blocks Files Data eompression Serviee routines 41 a string is sent indicating that the transmission failed and communications are termi- nated. Actual data characters are sent in blocks, which are smaller units of data that allow the data to be.checked as it is received. Since packets can be any size up to 64,000 bytes (this is expandable if future needs require), it is necessary to verify that communica- tions are proceeding successfully within the packet. This is done with data blocks. Blocks are smaller (currently 1,024 hyte) units of data sent within each packet. A block consists of a header and actual data characters. The header is a string identifying the beginning of a block and an integer indicating the number of data bytes contained in the block. The data is simply a string of characters which comprise data elements. Separate elements are delimited by a newline character. After the receiving end has received the number of characters specified in the block header, it sends a verification string to the sender indicating that the transmission was successful. The sending end then transmits the next block. If the receiving end does not get the proper number of characters in a specified amount of time, it sends a failure string indicating that the transmission was not completed. The sending end then re-transmits the previous block. If a single block fails repeatedly, transmission is terminated. The transfer of entire files from the UNIX server to the PC application uses a seperate mechanism built on top of the Packet transfer mechanism. Files are created on the server, broken into fixed sized packets and transfered one at a time to the PC client, where they are reassemhled into a copy of the original file. To support the transfer of point flow output files, three additional routines have been added to the server. These routines are designed to create the point flow output file, transfer the next section of the output file, and delete the server version of the file. The routines work in combination to create a temporary file on the server, transfer the file to the PC, and finally delete the server copy. The current version of SPWRMS does not use data compression. Previous versions utilized a Huffman encoding algorithm to reduce the number of bytes transmitted through the modems. However, this system proved incompatible with the terminal server at the Denver Office because personnel in that office preferred to use a local switch to enable configuration of the terminal server session. The compressed charac- ters would occasionally trigger the local switch. If data compression is re-implemented in the future, it is important to consider the configuration of the modems used for communication, as well as the other users who will be accessing the same modems. The service routines in the client-server arrangement are the functions that package and unpackage the data before and after it is transmitted. These routines basically read information out of the data structures, convert it to ASCII characters, and add the characters to the data string to be transmitted. It is imperative that the service routines on the PC and on the UNIX server application be identical, because there are built-in assumptions concerning the order in which data items are read from the data struc- tures. Any changes in the service routines on one platform must be echoed on the other platform.