This is the mail archive of the cygwin-developers@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

fifotest [45Kb]


I've finished my proof-of-concept code. The attached tarball (Sorry if
45kb is to much for you) is the testpack...

For this I need testers on windows 95,98,Me and NT 4 (As usual, it's
written from the msdn specs so it _should_ work :])

to build
./configure
the packages need the win32 API. if you have the current gcc no special
switchs are needed.

To test, fire up a few bash shells.
in one or more run ./fifoclient
in one or more fun ./fifotest.

The testpack is very screen intensive - it takes about 3 mins to run for
me with 1 writer, three readers.

suggested things to try:

a)start a few readers, then a writer, then kill the writer (the readers
should all stop)
b) the reverse.

c) start a few readers, then two writers then kill one, (nothing should
change)
d) start the writer again and kill the other (nothing should change)
e) the reverse.

If something abnormal happens above, please describe it clearly, and
include a screen shot showing the output of each process in your
feedback.

===
The test packages implement what will go into cygwin1.dll directly. If
this works on all platforms all I have to do to get a core
implementation running is migrate the code into the fifo handler I've
built, and implement the large data handling code. For that I need to
run some more testcases on unix to see the boundary behaviour...

Rob

fifotest-0.0.1.tar.gz


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]