On Tuesday 20 July 2004 17:17, Kobi Cohen-Arazi wrote:
> It is *much* faster than using semaphores, I've checked it. And yes - I
> could hardly believe it myself ;-)
If you talk about SysV semaphores (semget et-al), than they are
known to be quite expensive (a lot of extra funcionality, permission
Hey a friend translates it as the last idiot...
anyways another friend installed XP on five partitions on the same
computer and couldn't understand why nothing worked.
I don't know many who install their would OS, and partitioning under
linux I find to be much easier than with windoze.
Also the
Quoting Kobi Cohen-Arazi <[EMAIL PROTECTED]>:
> Hi,
>
> The test created consumer Thread Pool which were "sitting" on a
> synchronized message queue (using condition variables with mutex etc
> ...)
>
> The producer thread insert messages to that queue.
> The test measured the time took the system
On Tue, 2004-07-20 at 17:12, [EMAIL PROTECTED] wrote:
> Again we are talking about threads. The are no need to copy
> data to kernel, and than copy it back. So it will bot be faster at all.
It is *much* faster than using semaphores, I've checked it. And yes - I
could hardly believe it myself ;-)
Quoting Kobi Cohen-Arazi <[EMAIL PROTECTED]>:
The original design has some drawbacks. Fist of all
u need to understand that when threads are used, there are no
need to establish a solution that will suite to forked
processes.
You just need to create some class that will implement a queue.
Access
On Tue, 2004-07-20 at 15:18, Shachar Shemesh wrote:
> Dvir Volk wrote:
>
> > fifo's must be thread safe, because they support communication between
> > a number of processes.
> > but if all the threads are in the same process, and there's so little
> > data per read/write, why not use a queue in
Hi Shachar,
Have you considered using Unix domain sockets with DGRAM ?
Unix domain sockets as far as I understand, will be reliable even with
DGRAM, which allow you to benefit from the message oriented protocol.
There is only one gotchas here, the enqueuing threads need to make sure
they can queu
Hi all,
How atomic are read and write?
Background:
I want a queue implemented between different *threads*. At the moment,
there are several threads that need to send data, and only one thread
that retrieves the data from the queue. The most obvious answer seemed
to be "use a pipe".
Here's the s
Well, I was doing something along those lines until I moved to using
l2tpd, which works far better IMO. I have a script in my init.d,
called cables - it runs after l2tpd and before all other internet
requiring scripts. Took some adjustments of the runlevels :-) I've
attached the script, if anyone w
Hi,
Does anyone knows what these mean? and how to minimize them.
there is one line for every few minutes.
background: I have a cables connection to netvision.
pptp 1.1.0
pppd 2.4.1b2
kernel 2.4.4
connection line:
/usr/sbin/pptp-linux cable.netvision.net.il debug user $USERNAME remotename
cable.netv
Hi list.
I maintain a samba file server for a some company that uses it for
sharing database files for their windows based administration program.
Currently we experience many problems with a program named "Business
Organizer" that based on MS access. The program often crashes with error
messages.
Yes, I have used PayPal in the past. However, the downside to using
PayPal is that everyone using it must have a PayPal account and you exit
your site in order to do the processing. The client I am investigating
this for wants to avoid those issues so that they can have a customized
form for
12 matches
Mail list logo
