266. Random Access vs. Sequential Files

Up: Consistency and Semantics Next: Progress Previous: File Consistency

MPI distinguishes ordinary random access files from sequential stream files, such as pipes and tape files. Sequential stream files must be opened with the MPI_MODE_SEQUENTIAL flag set in the amode. For these files, the only permitted data access operations are shared file pointer reads and writes. Filetypes and etypes with holes are erroneous. In addition, the notion of file pointer is not meaningful; therefore, calls to MPI_FILE_SEEK_SHARED and MPI_FILE_GET_POSITION_SHARED are erroneous, and the pointer update rules specified for the data access routines do not apply. The amount of data accessed by a data access operation will be the amount requested unless the end of file is reached or an error is raised.


This implies that reading on a pipe will always wait until the requested amount of data is available or until the process writing to the pipe has issued an end of file. ( End of rationale.)
Finally, for some sequential files, such as those corresponding to magnetic tapes or streaming network connections, writes to the file may be destructive. In other words, a write may act as a truncate (a MPI_FILE_SET_SIZE with size set to the current position) followed by the write.

Up: Consistency and Semantics Next: Progress Previous: File Consistency

Return to MPI-2.1 Standard Index
Return to MPI Forum Home Page

MPI-2.0 of July 1, 2008
HTML Generated on July 6, 2008