m0p@k.cc.purdue.edu (S. Kulikowski) (09/22/88)
need debugging help. boy, this is a wierd one and i am at wits end.
i have written some realtime programs which consult the MSDOS clock
frequently and record performance data in files. running on a DEC rainbow
with MSDOS v. 2.11.1. the program with the bug is written in Turbo
Pascal and i have compiled it with TP3 and TP4 and the bug survives both.
last year we ran 70+ kids through the program in a research project and
each kid produced 8 files each with data in them. some 40+ megabytes in
total data. upon analysis, 10 files were found to be truncated.
the only thing they have in common was that they had been operating at
9:06 AM when no further data entered the files beyond the time 9:06:09 AM.
research notes and subsequent testing show that the program IO continues
on the monitor as usual but NO FURTHER DATA ENTERS THE FILE AFTER 9:06:09.
this happens in any location (3 different schools and in our lab at the
university). all other times (so far) appear to be rock solid.
as a kludge, i can have the RAs set the clock to 9:07 in the morning
sessions before starting, but i am queasy about a data program which will
putz out for unexplained reasons.
any ideas on how to debug it? what could be happening at 9:06 AM that
could cause file saving to go bye-bye? i am going to recompile the
program on an IBM PC and see if the problem is DEC specific, but this
would essentially cost us our research machines.
any help would be great appreciated,
stan
BITNET : XM0P @ PURCCVM (* note, zero, not Oh *)
SnailMail : Special Education; Purdue University; W. Lafayette, IN 47907
USENET : k.cc.purdue.edu!m0p COMPUSERVE : 75410,1211