Search Results

Search found 5456 results on 219 pages for 'named pipes'.

Page 20/219 | < Previous Page | 16 17 18 19 20 21 22 23 24 25 26 27  | Next Page >

  • Optional Parameters and Named Arguments in C# 4 (and a cool scenario w/ ASP.NET MVC 2)

    [In addition to blogging, I am also now using Twitter for quick updates and to share links. Follow me at: twitter.com/scottgu] This is the seventeenth in a series of blog posts Im doing on the upcoming VS 2010 and .NET 4 release. Todays post covers two new language feature being added to C# 4.0 optional parameters and named arguments as well as a cool way you can take advantage of optional parameters (both in VB and C#) with ASP.NET MVC 2. Optional Parameters in C# 4.0 C# 4.0 now...Did you know that DotNetSlackers also publishes .net articles written by top known .net Authors? We already have over 80 articles in several categories including Silverlight. Take a look: here.

    Read the article

  • JPQL IN clause: Java-Arrays (or Lists, Sets...)?

    - by Bernd Haug
    I would like to load all objects that have a textual tag set to any of a small but arbitrary number of values from our database. The logical way to go about this in SQL would be to build an "IN" clause. JPQL allows for IN, but it seems to require me to specify every single parameter to IN directly (as in, "in (:in1, :in2, :in3)"). Is there some way to specify an array, or a list (or some other container) that should be unrolled to the values of an IN clause?

    Read the article

  • Anchor tag not centered on page after clicking

    - by jennyjune
    I am creating a website that scrolls down to different divs by the use of anchors. However, when I click on a link, that anchored div is not centered on the page. I positioned my divs absolutely but I have no idea if that is affecting it? Anyone have any clue how to get the divs centered using anchors? Please help! <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> <html xmlns="http://www.w3.org/1999/xhtml"> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> <title>Jean and James</title> <script src="js/jquery.js" type="text/javascript"></script> <script src="js/jquery.anchor.js" type="text/javascript"></script> <link href="jj_css.css" rel="stylesheet" type="text/css" /> </head> <body> <div id="navigation" align="center"> <a href="#jeanandjames" class="anchorLink">JEAN AND JAMES</a> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <a href="#feature" class="anchorLink">FEATURE PRODUCT</a> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <a href="#contact" class="anchorLink">CONTACT</a> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <a href="#work" class="anchorLink">WORK</a> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </div> <div id="logo"> <img src="logo.gif" /> </div> <a name="jeanandjames" id="jeanandjames"> <div id="jeanandjames" align="center"> <img src="jeanandjames.jpg" /><br /> A boutique design firm specializing in hand crafted products and graphic print work. </div> </a> <a name="feature" id="feature"> <div id="feature"> <img src="baby_bowtie.gif" /> </div> </a> <a name="contact" id="contact"> <div id="contact"> Jean & James Chicago, IL<br /><br /> Donec faucibus nunc eget quam pretium vel porttitor diam venenatis. Nullam egestas, quam et congue rutrum, diam mauris tempus urna, eget mattis quam quam ac erat. Vivamus lobortis nunc quis arcu lobortis vel euismod felis tincidunt. Etiam facilisis, metus et mattis tempor, erat ipsum tincidunt diam, nec aliquam velit risus a lacus. Duis tempor bibendum felis, adipiscing mattis lorem eleifend fermentum. Nulla adipiscing consequat enim id sagittis. Suspendisse venenatis scelerisque orci, sed sollicitudin erat egestas in. Nunc orci leo, adipiscing vel consectetur sit amet, rutrum vitae nulla. Mauris ut mi nunc, non condimentum risus. Donec varius consequat quam vel lacinia. Etiam et eros nec quam egestas pellentesque sit amet ac nunc. </div> </a> </body> </html> And here is my css: #navigation A:link {text-decoration: none; color: #FFFFFF;} #navigation A:visited {text-decoration: none; color: #FFFFFF;} #navigation A:active {text-decoration: none; color: #FFFFFF;} #navigation A:hover {text-decoration: none; color: #CC9;} #navigation { position: fixed; top: -1px; left: -1px; width: 100%; height: 20px; padding: 10px; z-index: 10; font-family: Arial, Helvetica, sans-serif; font-size: 12px; background-color: #000; color: #FFF; } #logo { position: fixed; top: 150px; left: 50px; } #jeanandjames { position: absolute; top: 230px; left: 200px; font-family: Arial, Helvetica, sans-serif; font-size: 12px; } #feature { position: absolute; top: 600px; left: 200px; } #contact { position: absolute; top: 1200px; left: 200px; width: 500px; font-family: Arial, Helvetica, sans-serif; font-size: 12px; }

    Read the article

  • will_paginate with named_scopes

    - by Lowgain
    I'm using will_paginate for pagination, which has been working well so far, except for this one thing. If I try to paginate a scope, for instance class User < ActiveRecord::Base named_scope :scope, lambda { etc } end User.scope.paginate({:page => params[:page], :per_page => 10}) That will tell me paginate is an undefined method. I'd rather not have to use a second solution for only this scope, is there something I can do here?

    Read the article

  • Alerting for incorrect cell values

    - by Allan
    My problem: I have two ranges R16 and R01. These ranges were set up by swiping each range and then renaming them in the upper left panel of the sheet. Each range requires users to fill in each cell with a value. R16 requires users to enter a number of 0 through 5. The range R01 requires a value of 0 or 1 to be entered. NO cell can be left blank in any cell within these two ranges. These ranges and requirements are specific to this sheet only. It would be nice if at the time of user entering a number, an error message appeared like [invalid entry] if the value inputted was outside parameters set. For example, in R16, if someone entered 12 or -1 they would be alerted. Finally when the user presses a button on the page to use these values in a separate process, it is essential to check that no cell is left blank. I am trying to find a way to halt the running of the marco (via the button) if these parameters above are not met. Thank you

    Read the article

  • named_scope or find_by_sql?

    - by keruilin
    I have three models: User Award Trophy The associations are: User has many awards Trophy has many awards Award belongs to user Award belongs to trophy User has many trophies through awards Therefore, user_id is a fk in awards, and trophy_id is a fk in awards. In the Trophy model, which is an STI model, there's a trophy_type column. I want to return a list of users who have been awarded a specific trophy -- (trophy_type = 'GoldTrophy'). Users can be awarded the same trophy more than once. (I don't want distinct results.) Can I use a named_scope? How about chaining them? Or do I need to use find_by_sql? Either way, how would I code it?

    Read the article

  • How transform this find_by_sql to named_scope?

    - by keruilin
    How can I possibly turn into named_scope? def self.hero_badge_awardees return User.find_by_sql("select users.*, awards.*, badges.badge_type from users, awards, badges where awards.user_id = users.id and badges.id = awards.badge_id and badges.badge_type = 'HeroBadge'") end

    Read the article

  • How to filter results by multiple fields?

    - by hadees
    I am working on a survey application in ruby on rails and on the results page I want to let users filter the answers by a bunch of demographic questions I asked at the start of the survey. For example I asked users what their gender and career was. So I was thinking of having dropdowns for gender and career. Both dropdowns would default to all but if a user selected female and marketer then my results page would so only answers from female marketers. I think the right way of doing this is to use named_scopes where I have a named_scope for every one of my demographic questions, in this example gender and career, which would take in a sanitized value from the dropdown to use at the conditional but i'm unsure on how to dynamically create the named_scope chain since I have like 5 demographic questions and presumably some of them are going to be set to all.

    Read the article

  • rails named_scope issue with eager loading

    - by Craig
    Two models (Rails 2.3.8): User; username & disabled properties; User has_one :profile Profile; full_name & hidden properties I am trying to create a named_scope that eliminate the disabled=1 and hidden=1 User-Profiles. Moreover, while the User model is usually used in conjunction with the Profile model, I would like the flexibility to be able specify this using the :include = :profile syntax. I have the following User named_scope: named_scope :visible, { :joins => "INNER JOIN profiles ON users.id=profiles.user_id", :conditions => ["users.disabled = ? AND profiles.hidden = ?", false, false] } This works as expected when just reference the User model: >> User.visible.map(&:username).flatten => ["user a", "user b", "user c", "user d"] However, when I attempt to include the Profile model: User.visible(:include=> :profiles).profile.map(&:full_name).flatten I get an error that reads: NoMethodError: undefined method `profile' for #<User:0x1030bc828> Am I able to cross model-collection boundaries in this manner?

    Read the article

  • Design: How to declare a specialized memory handler class

    - by Michael Dorgan
    On an embedded type system, I have created a Small Object Allocator that piggy backs on top of a standard memory allocation system. This allocator is a Boost::simple_segregated_storage< class and it does exactly what I need - O(1) alloc/dealloc time on small objects at the cost of a touch of internal fragmentation. My question is how best to declare it. Right now, it's scope static declared in our mem code module, which is probably fine, but it feels a bit exposed there and is also now linked to that module forever. Normally, I declare it as a monostate or a singleton, but this uses the dynamic memory allocator (where this is located.) Furthermore, our dynamic memory allocator is being initialized and used before static object initialization occurs on our system (as again, the memory manager is pretty much the most fundamental component of an engine.) To get around this catch 22, I added an extra 'if the small memory allocator exists' to see if the small object allocator exists yet. That if that now must be run on every small object allocation. In the scheme of things, this is nearly negligable, but it still bothers me. So the question is, is there a better way to declare this portion of the memory manager that helps decouple it from the memory module and perhaps not costing that extra isinitialized() if statement? If this method uses dynamic memory, please explain how to get around lack of initialization of the small object portion of the manager.

    Read the article

  • Translate SQL statement into named_scope?

    - by keruilin
    How can I translate this SQL into a named_scope? Also, I want the total comments param to be passed through a lambda. "select users., count() as total_comments from users, comments where (users.id = comments.user_id) and (comments.public_comment = 1) and (comments.aasm_state = 'posted') and (comments.forum_user_id is null) group by users.id having total_comments 25"

    Read the article

  • Reusing named_scope to define another named_scope

    - by Sergei Kozlov
    The problem essence as I see it One day, if I'm not mistaken, I have seen an example of reusing a named_scope to define another named_scope. Something like this (can't remember the exact syntax, but that's exactly my question): named_scope :billable, :conditions => ... named_scope :billable_by_tom, :conditions => { :billable => true, :user => User.find_by_name('Tom') } The question is: what is the exact syntax, if it's possible at all? I can't find it back, and Google was of no help either. Some explanations Why I actually want it, is that I'm using Searchlogic to define a complex search, which can result in an expression like this: Card.user_group_managers_salary_greater_than(100) But it's too long to be put everywhere. Because, as far as I know, Searchlogic simply defines named_scopes on the fly, I would like to set a named_scope on the Card class like this: named_scope from_big_guys, { user_group_managers_salary_greater_than(100) } - this is where I would use that long Searchlogic method inside my named_scope. But, again, what would be the syntax? Can't figure it out. Resume So, is named_scope nesting (and I do not mean chaining) actually possible?

    Read the article

  • How to catch-up named mercurial branch from default branch without merging the two into one?

    - by Dynite
    I have two branches in mercurial.. default named |r1 |r2 |r3 -------- named branch created here. | |r4 | |r5 | r6 | | |r7 | | -----------> | r8 How do I achieve this catch-up? | | I want to update the named branch from default, but I'm not ready to merge the branches yet. How do I achieve this? Edit: Additionally, what would the operation be using the GUI? Is it.. right-click r6, merge with..., r8,... then what? commit to named branch?

    Read the article

  • How do I use a named_scope to filter records in my model

    - by kibyegon
    I have a model "Product" with a "description" field. Now I want to have a link in the index page that when clicked will show all products where the description is blank (empty). In the model I have defined a named_scope like this named_scope :no_description, :conditions => { :description => "" } I have checked that the named_scope works by calling Product.no_description.count on the console. As far as I know, the controller is then supposed to handle the filter request from the link on the "index" action but be able to distinguish it from the default which is view all products. def index @products = Product.all ... My problem is getting the controller handle the different request, what route to setup for the link on the view and the actual link on the view. Hope I explained my problem.

    Read the article

  • What does ‘?’ stand for in SQL?

    - by user295189
    I have this SQL by a programmer: $sql = " INSERT INTO `{$database}`.`table` ( `my_id`, `xType`, `subType`, `recordID`, `textarea` ) VALUES ( {$my_id}, ?xType, ?subType, {$recordID}, ?areaText ) "; My question is why is he using ? before values? How do I see what values are coming in? I did echo and it shows ?xType as ?xType. No values. What does ? stand for in SQL?

    Read the article

  • Should I Solve this with Multithreading in Ruby?

    - by viatropos
    I have a strange case, here's the sequence of actions: User edits a document and hits save Application sends GET request to service Service sends POST request back to application in the middle of responding to the GET request Application, in the same state as when it made the GET request, responds to the POST request (sends document data) to service. Service sends data back to Application (responding to original GET request) Application handles the rest... The use case is this: I was thinking how can I make Yahoo Pipes POST data? Specifically, I want it to be able to update Google Docs when a user makes a change locally (on a custom editor). So user edits doc, makes GET request to Yahoo Pipes, Pipes makes a POST request back to App to get the document (Pipes can only make this type of POST request), App sends doc, Pipes formats data according to the Google API, Pipes responds to GET request with Google API formatted XML, App makes the post request. Theoretically, how would I accomplish this? It seems that I need to create a separate ruby Process for the GET request, and when Pipes sends the POST request, I find that process and send its output, then I'm stuck. This would cut out the need for a database for this particular case (I could save the stuff temporarily in a database, but that doesn't seem right). Any ideas? This would make it so I don't have to format things to the Google API in ruby, I could leave that to Pipes.

    Read the article

  • Enabling DNS for IPv6 infrastructure

    After successful automatic distribution of IPv6 address information via DHCPv6 in your local network it might be time to start offering some more services. Usually, we would use host names in order to communicate with other machines instead of their bare IPv6 addresses. During the following paragraphs we are going to enable our own DNS name server with IPv6 address resolving. This is the third article in a series on IPv6 configuration: Configure IPv6 on your Linux system DHCPv6: Provide IPv6 information in your local network Enabling DNS for IPv6 infrastructure Accessing your web server via IPv6 Piece of advice: This is based on my findings on the internet while reading other people's helpful articles and going through a couple of man-pages on my local system. What's your name and your IPv6 address? $ sudo service bind9 status * bind9 is running If the service is not recognised, you have to install it first on your system. This is done very easy and quickly like so: $ sudo apt-get install bind9 Once again, there is no specialised package for IPv6. Just the regular application is good to go. But of course, it is necessary to enable IPv6 binding in the options. Let's fire up a text editor and modify the configuration file. $ sudo nano /etc/bind/named.conf.optionsacl iosnet {        127.0.0.1;        192.168.1.0/24;        ::1/128;        2001:db8:bad:a55::/64;};listen-on { iosnet; };listen-on-v6 { any; };allow-query { iosnet; };allow-transfer { iosnet; }; Most important directive is the listen-on-v6. This will enable your named to bind to your IPv6 addresses specified on your system. Easiest is to specify any as value, and named will bind to all available IPv6 addresses during start. More details and explanations are found in the man-pages of named.conf. Save the file and restart the named service. As usual, check your log files and correct your configuration in case of any logged error messages. Using the netstat command you can validate whether the service is running and to which IP and IPv6 addresses it is bound to, like so: $ sudo service bind9 restart $ sudo netstat -lnptu | grep "named\W*$"tcp        0      0 192.168.1.2:53        0.0.0.0:*               LISTEN      1734/named      tcp        0      0 127.0.0.1:53          0.0.0.0:*               LISTEN      1734/named      tcp6       0      0 :::53                 :::*                    LISTEN      1734/named      udp        0      0 192.168.1.2:53        0.0.0.0:*                           1734/named      udp        0      0 127.0.0.1:53          0.0.0.0:*                           1734/named      udp6       0      0 :::53                 :::*                                1734/named   Sweet! Okay, now it's about time to resolve host names and their assigned IPv6 addresses using our own DNS name server. $ host -t aaaa www.6bone.net 2001:db8:bad:a55::2Using domain server:Name: 2001:db8:bad:a55::2Address: 2001:db8:bad:a55::2#53Aliases: www.6bone.net is an alias for 6bone.net.6bone.net has IPv6 address 2001:5c0:1000:10::2 Alright, our newly configured BIND named is fully operational. Eventually, you might be more familiar with the dig command. Here is the same kind of IPv6 host name resolve but it will provide more details about that particular host as well as the domain in general. $ dig @2001:db8:bad:a55::2 www.6bone.net. AAAA More details on the Berkeley Internet Name Domain (bind) daemon and IPv6 are available in Chapter 22.1 of Peter Bieringer's HOWTO on IPv6. Setting up your own DNS zone Now, that we have an operational named in place, it's about time to implement and configure our own host names and IPv6 address resolving. The general approach is to create your own zone database below the bind folder and to add AAAA records for your hosts. In order to achieve this, we have to define the zone first in the configuration file named.conf.local. $ sudo nano /etc/bind/named.conf.local //// Do any local configuration here//zone "ios.mu" {        type master;        file "/etc/bind/zones/db.ios.mu";}; Here we specify the location of our zone database file. Next, we are going to create it and add our host names, our IP and our IPv6 addresses. $ sudo nano /etc/bind/zones/db.ios.mu $ORIGIN .$TTL 259200     ; 3 daysios.mu                  IN SOA  ios.mu. hostmaster.ios.mu. (                                2014031101 ; serial                                28800      ; refresh (8 hours)                                7200       ; retry (2 hours)                                604800     ; expire (1 week)                                86400      ; minimum (1 day)                                )                        NS      server.ios.mu.$ORIGIN ios.mu.server                  A       192.168.1.2server                  AAAA    2001:db8:bad:a55::2client1                 A       192.168.1.3client1                 AAAA    2001:db8:bad:a55::3client2                 A       192.168.1.4client2                 AAAA    2001:db8:bad:a55::4 With a couple of machines in place, it's time to reload that new configuration. Note: Each time you are going to change your zone databases you have to modify the serial information, too. Named loads the plain text zone definitions and converts them into an internal, indexed binary format to improve lookup performance. If you forget to change your serial then named will not use the new records from the text file but the indexed ones. Or you have to flush the index and force a reload of the zone. This can be done easily by either restarting the named: $ sudo service bind9 restart or by reloading the configuration file using the name server control utility - rndc: $ sudo rndc reconfig Check your log files for any error messages and whether the new zone database has been accepted. Next, we are going to resolve a host name trying to get its IPv6 address like so: $ host -t aaaa server.ios.mu. 2001:db8:bad:a55::2Using domain server:Name: 2001:db8:bad:a55::2Address: 2001:db8:bad:a55::2#53Aliases: server.ios.mu has IPv6 address 2001:db8:bad:a55::2 Looks good. Alternatively, you could have just ping'd the system as well using the ping6 command instead of the regular ping: $ ping6 serverPING server(2001:db8:bad:a55::2) 56 data bytes64 bytes from 2001:db8:bad:a55::2: icmp_seq=1 ttl=64 time=0.615 ms64 bytes from 2001:db8:bad:a55::2: icmp_seq=2 ttl=64 time=0.407 ms^C--- ios1 ping statistics ---2 packets transmitted, 2 received, 0% packet loss, time 1001msrtt min/avg/max/mdev = 0.407/0.511/0.615/0.104 ms That also looks promising to me. How about your configuration? Next, it might be interesting to extend the range of available services on the network. One essential service would be to have web sites at hand.

    Read the article

  • How to configure DNS Server on Fedora

    - by user863873
    I want to learn how to configure my home PC server into a web server with domain and host. My IP is 109.99.141.133 and now points to a phpinfo page host on my home server. My registed domain is: anunta-anunturi.ro I searched for a tutorial and I've read that I have to configure /etc/named.conf and the file sources for the new zone that I create. So, from the tutorials, my /etc/named.conf looks like this: // // named.conf // // Provided by Red Hat bind package to configure the ISC BIND named(8) DNS // server as a caching only nameserver (as a localhost DNS resolver only). // // See /usr/share/doc/bind*/sample/ for example named configuration files. // options { listen-on port 53 { 127.0.0.1; }; listen-on-v6 port 53 { ::1; }; directory "/var/named"; dump-file "/var/named/data/cache_dump.db"; statistics-file "/var/named/data/named_stats.txt"; memstatistics-file "/var/named/data/named_mem_stats.txt"; allow-query { localhost; }; recursion yes; dnssec-enable yes; dnssec-validation yes; dnssec-lookaside auto; /* Path to ISC DLV key */ bindkeys-file "/etc/named.iscdlv.key"; managed-keys-directory "/var/named/dynamic"; }; logging { channel default_debug { file "data/named.run"; severity dynamic; }; }; zone "anunta-anunturi.ro" IN { type master; file "/etc/anunta-anunturi.db"; }; zone "." IN { type hint; file "named.ca"; }; include "/etc/named.rfc1912.zones"; include "/etc/named.root.key"; My /etc/anunta-anunturi.db file looks like this — I'm not sure if this is okay, or if it's the easy one. $TTL 86400 anunta-anunturi.ro. IN SOA serveur.anunta-anunturi.ro. root.serveur.anunta-anunturi.ro. ( 1997022700 ; Serial 28800 ; Refresh 14400 ; Retry 3600000 ; Expire 86400 ) ; Minumun IN NS serveur.anunta-anunturi.ro. IN MX 10 mail.anunta-anunturi.ro. serveur.anunta-anunturi.ro. IN A 192.168.1.37 www.anunta-anunturi.ro. IN A 192.168.1.37 mail.anunta-anunturi.ro. IN A 192.168.1.37 Extra info: At home I receive internet from my ISP through a router. My home PC and server recieve their IP automatically from the router when I start/restart. In my local home network, my server receives the IP 192.168.1.37 from the router. When I enter 109.99.141.133 in my browser, it points to the rooter that forwards port 80 to local IP 192.168.1.37 (my home server) Questions: Are my two files good? What/where is my nameserver that I need to copy/paste to my top level domain (where I registered my domain: rotld.ro)?

    Read the article

  • Why do these ipfw delayed pipes have no effect?

    - by troutwine
    I'm on OSX 10.7.5 and am attempting to add some latency to the connection to my personal domain with ipfw, using this article as a guide. Normal latency: > ping -c5 troutwine.us PING troutwine.us (198.101.227.131): 56 data bytes 64 bytes from 198.101.227.131: icmp_seq=0 ttl=56 time=92.714 ms 64 bytes from 198.101.227.131: icmp_seq=1 ttl=56 time=91.436 ms 64 bytes from 198.101.227.131: icmp_seq=2 ttl=56 time=91.218 ms 64 bytes from 198.101.227.131: icmp_seq=3 ttl=56 time=91.451 ms 64 bytes from 198.101.227.131: icmp_seq=4 ttl=56 time=91.243 ms --- troutwine.us ping statistics --- 5 packets transmitted, 5 packets received, 0.0% packet loss round-trip min/avg/max/stddev = 91.218/91.612/92.714/0.559 ms Enabling ipfw: > sudo sysctl -w net.inet.ip.fw.enable=0 net.inet.ip.fw.enable: 1 -> 0 > sudo sysctl -w net.inet.ip.fw.enable=1 net.inet.ip.fw.enable: 0 -> 1 The configuration of the pipe: > sudo ipfw add pipe 1 ip from any to 198.101.227.131 00200 pipe 1 ip from any to any dst-ip 198.101.227.131 > sudo ipfw add pipe 2 ip from 198.101.227.131 to any 00500 pipe 2 ip from 198.101.227.131 to any > sudo ipfw pipe 1 config delay 250ms bw 1Mbit/s plr 0.1 > sudo ipfw pipe 2 config delay 250ms bw 1Mbit/s plr 0.1 The pipes are in place and configured: > sudo ipfw -a list 00100 166 14178 fwd 127.0.0.1,20559 tcp from any to me dst-port 80 in 00200 0 0 pipe 1 ip from any to 198.101.227.131 00300 0 0 pipe 2 ip from 198.101.227.131 to any 65535 37452525 32060610029 allow ip from any to any > sudo ipfw pipe list 00001: 1.000 Mbit/s 250 ms 50 sl.plr 0.100000 0 queues (1 buckets) droptail mask: 0x00 0x00000000/0x0000 -> 0x00000000/0x0000 00002: 1.000 Mbit/s 250 ms 50 sl.plr 0.100000 0 queues (1 buckets) droptail mask: 0x00 0x00000000/0x0000 -> 0x00000000/0x0000 Yet, this has had no effect: > ping -c5 troutwine.us PING troutwine.us (198.101.227.131): 56 data bytes 64 bytes from 198.101.227.131: icmp_seq=0 ttl=56 time=100.920 ms 64 bytes from 198.101.227.131: icmp_seq=1 ttl=56 time=91.648 ms 64 bytes from 198.101.227.131: icmp_seq=2 ttl=56 time=91.777 ms 64 bytes from 198.101.227.131: icmp_seq=3 ttl=56 time=91.466 ms 64 bytes from 198.101.227.131: icmp_seq=4 ttl=56 time=93.209 ms --- troutwine.us ping statistics --- 5 packets transmitted, 5 packets received, 0.0% packet loss round-trip min/avg/max/stddev = 91.466/93.804/100.920/3.612 ms What gives? I understand that ipfw is depreciated, but the manpage does not mention it being disabled. Also, I am not using Network Link Controller as I want to affect a single host.

    Read the article

  • What's the difference between the BitTorrent clients named "BitTorrent" and "µTorrent"?

    - by Eric
    A similar question was asked but never really addressed the question, I think in part because of terminology confusion. So to be very clear: what's the difference between the two BitTorrent client, one named "BitTorrent" and the one named "µTorrent"? They look to have identical UIs, right down to the same checkboxes in the prefs dialogs. Why are there two programs with different names that appear to be identical? Is one superior to the other? (Are they different in any way?) Thanks.

    Read the article

  • Reloading Apache httpd on Windows - error 'No installed service named "Apache2.2".'

    - by user143228
    This is a variant of How do I restart Apache on Windows? "Apache -k restart" gives error "No installed service named "Apache2"; I have Apache httpd 2.2.22 on Windows, not running as a service (our product's uber-service starts httpd as a console app). I'm trying to get httpd to reload its configuration; Apache's Windows docs suggest httpd -k restart from any console window. When I do that, I get PS C:\apache\bin> .\httpd.exe -k restart [Mon Oct 29 14:06:56 2012] [error] (OS 2)The system cannot find the file specified. : No installed service named "Apache2.2". How do I convince it Apache's not running as a service?

    Read the article

  • Debian 5 server is randomly shutting down.

    - by revofreak
    My debian 5 vps is suffering from random shutdowns. I reinstalled it several times, the hosts moved me to a different physical box, check the install image and said everyone else also uses it and is fine. Heres the output from syslog Mar 27 00:19:19 noobintraining-1 -- MARK -- Mar 27 00:32:01 noobintraining-1 shutdown[18142]: shutting down for system halt Mar 27 00:32:06 noobintraining-1 init: Switching to runlevel: 0 Mar 27 00:32:06 noobintraining-1 xinetd[15907]: Exiting... Mar 27 00:32:07 noobintraining-1 named[15865]: received control channel command 'stop -p' Mar 27 00:32:07 noobintraining-1 named[15865]: shutting down: flushing changes Mar 27 00:32:07 noobintraining-1 named[15865]: stopping command channel on 127.0.0.1#953 Mar 27 00:32:07 noobintraining-1 named[15865]: stopping command channel on ::1#953 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on ::#53 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on 127.0.0.1#53 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on 89.238.172.132#53 Mar 27 00:32:07 noobintraining-1 named[15865]: exiting Mar 27 00:32:07 noobintraining-1 exiting on signal 15 Any help is most appreciated!

    Read the article

  • Debian 5 is randomly shutting down.

    - by revofreak
    My debian 5 vps is suffering from random shutdowns. I reinstalled it several times, the hosts moved me to a different physical box, check the install image and said everyone else also uses it and is fine. Heres the output from syslog Mar 27 00:19:19 noobintraining-1 -- MARK -- Mar 27 00:32:01 noobintraining-1 shutdown[18142]: shutting down for system halt Mar 27 00:32:06 noobintraining-1 init: Switching to runlevel: 0 Mar 27 00:32:06 noobintraining-1 xinetd[15907]: Exiting... Mar 27 00:32:07 noobintraining-1 named[15865]: received control channel command 'stop -p' Mar 27 00:32:07 noobintraining-1 named[15865]: shutting down: flushing changes Mar 27 00:32:07 noobintraining-1 named[15865]: stopping command channel on 127.0.0.1#953 Mar 27 00:32:07 noobintraining-1 named[15865]: stopping command channel on ::1#953 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on ::#53 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on 127.0.0.1#53 Mar 27 00:32:07 noobintraining-1 named[15865]: no longer listening on 89.238.172.132#53 Mar 27 00:32:07 noobintraining-1 named[15865]: exiting Mar 27 00:32:07 noobintraining-1 exiting on signal 15 Any help is most appreciated!

    Read the article

  • Would using a MemoryMappedFile for IPC across AppDomains be faster than WCF/named pipes?

    - by Morten Mertner
    Context: I am loading and executing untrusted code in a separate AppDomain and am currently communicating between the two using WCF (using named pipes as the underlying transport). I am exchanging relatively simple object graphs using a reasonably coarse-grained API, but would like to use a more fine-grained API if it does not cost me performance-wise. I've noticed that 4.0 adds a MemoryMappedFile class (which doesn't need a physical file, so could be entirely memory based). What kind of performance gains could I expect to see (if any) by using this new class? I know that it would take some "infrastructure code" to get the request/response behavior of WCF, but for now I'm only interested in the performance difference.

    Read the article

< Previous Page | 16 17 18 19 20 21 22 23 24 25 26 27  | Next Page >