monitoring the filesystem for changes

[rohit]

hello,
i have designed a desktop search utility in python and a file system
monitoring using readdirectorych angesw from win32api but for eg. it
has a high cpu utilization (using a 2GHz processor).
a solution ntfs change journal proposed here works only for ntfs
partitions
is there another low (processor) cost alternative that works on fat32
partitions as well.
Rohit

[momobear]

Re: monitoring the filesystem for changes

i have designed a desktop search utility in python and a file system
monitoring using readdirectorych angesw from win32api but for eg. it
has a high cpu utilization (using a 2GHz processor).

I don't think so. I just build a service to monitor a directory
changes using readdirectorych angesw, the cpu usage is not above 1.9%
for this service(my computer's cpu is 1.7GHz). if you still think 1.9%
is high cpu utilization, I think you can use the c to implements this
function, the cpu usage seems none for this application(I tried c dll
solution first).
Wish this help to you.

Wang Wei

[rohit]

Re: monitoring the filesystem for changes

hello
well in my implementation of readdirectorych angesw i am using
threading that is multiple programs run simultaneously each program
monitoring one drive for changes.
is their a way around this...one program for all drives?
thanks
rohit

[Tim Golden]

Re: monitoring the filesystem for changes

rohit wrote:

hello
well in my implementation of readdirectorych angesw i am using
threading that is multiple programs run simultaneously each program
monitoring one drive for changes.
is their a way around this...one program for all drives?
thanks

You could use the ReadDirectoryCh angesW in overlapped
mode. I've never tried it, so I don't know how
robust it would be. Why is it a problem to have
"multiple programs" running? And is that simply
multiple threads, or multiple processes?

TJG

[momobear]

Re: monitoring the filesystem for changes

You could use the ReadDirectoryCh angesW in overlapped

mode. I've never tried it, so I don't know how
robust it would be. Why is it a problem to have
"multiple programs" running? And is that simply
multiple threads, or multiple processes?

I used overlapped in my program, it runs about 1 week, seems no
problems now.
FILE_LIST_DIREC TORY = 0x0001
hDir = None

def __init__(self, directory):
self.hDir = win32file.Creat eFile (
directory,
self.FILE_LIST_ DIRECTORY,
win32con.FILE_S HARE_READ | win32con.FILE_S HARE_WRITE,
None,
win32con.OPEN_E XISTING,
win32con.FILE_F LAG_BACKUP_SEMA NTICS|
win32con.FILE_F LAG_OVERLAPPED,
None
)
but I didn't try use multiple program do the same action. I think only
one thread to do directory monitor is enough. other thread and process
can communicate with it use thread or process communication.

Wang Wei