Search Results

Search found 8224 results on 329 pages for 'sometimes'.

Page 232/329 | < Previous Page | 228 229 230 231 232 233 234 235 236 237 238 239  | Next Page >

  • Library to fake intermittent failures according to tester-defined policy?

    - by crosstalk
    I'm looking for a library that I can use to help mock a program component that works only intermittently - usually, it works fine, but sometimes it fails. For example, suppose I need to read data from a file, and my program has to avoid crashing or hanging when a read fails due to a disk head crash. I'd like to model that by having a mock data reader function that returns mock data 90% of the time, but hangs or returns garbage otherwise. Or, if I'm stress-testing my full program, I could turn on debugging code in my real data reader module to make it return real data 90% of the time and hang otherwise. Now, obviously, in this particular example I could just code up my mock manually to test against a random() routine. However, I was looking for a system that allows implementing any failure policy I want, including: Fail randomly 10% of the time Succeed 10 times, fail 4 times, repeat Fail semi-randomly, such that one failure tends to be followed by a burst of more failures Any policy the tester wants to define Furthermore, I'd like to be able to change the failure policy at runtime, using either code internal to the program under test, or external knobs or switches (though the latter can be implemented with the former). In pig-Java, I'd envision a FailureFaker interface like so: interface FailureFaker { /** Return true if and only if the mocked operation succeeded. Implementors should override this method with versions consistent with their failure policy. */ public boolean attempt(); } And each failure policy would be a class implementing FailureFaker; for example there would be a PatternFailureFaker that would succeed N times, then fail M times, then repeat, and a AlwaysFailFailureFaker that I'd use temporarily when I need to simulate, say, someone removing the external hard drive my data was on. The policy could then be used (and changed) in my mock object code like so: class MyMockComponent { FailureFaker faker; public void doSomething() { if (faker.attempt()) { // ... } else { throw new RuntimeException(); } } void setFailurePolicy (FailureFaker policy) { this.faker = policy; } } Now, this seems like something that would be part of a mocking library, so I wouldn't be surprised if it's been done before. (In fact, I got the idea from Steve Maguire's Writing Solid Code, where he discusses this exact idea on pages 228-231, saying that such facilities were common in Microsoft code of that early-90's era.) However, I'm only familiar with EasyMock and jMockit for Java, and neither AFAIK have this function, or something similar with different syntax. Hence, the question: Do such libraries as I've described above exist? If they do, where have you found them useful? If you haven't found them useful, why not?

    Read the article

  • Forms&Reports upgrade characterset issues

    - by Lukasz Romaszewski
    Hello,This quick post is based on my findings during recent IMC workshops, especially those related to upgrading the Forms 6i/9i/10g applications to Forms 11g platform. The upgrade process itself is pretty straightforward and it basically requires recompiling your Forms application with a latest version of frmcmp tool. For some cases though, especially when you migrate from Forms 6i which is a client-server architecture to a 3-tier web solution (Forms 11g), you need to rewrite some parts of your code to make it run on new platform. The things you need to change range from reimplementing (using webutil library) typical client-site functionality like local IO operation, access to WinAPI, invoking DLLs etc. to changing deprecated or obsolete APIs like RUN_PRODUCT to RUN_REPORT_OBJECT. To automate those changes Oracle provides complete Java API  which allows you to manipulate the code and structure of you modules (JDAPI). To make it even easier we can use Forms Migration Assistant tool (written in Java using JDAPI) which is able to replace all occurrences of old API entries with their 11g equivalents or warn you when the replacement is not possible. You can also add your own replacement definitions in the search_replace.properties file. But you need to be aware of some issues that can be encountered using this tool. First of all if you are using some hard-coded text inside your triggers you may notice that after processing them by the Migration Assistant tool the national characters may be lost. This is due to the fact that you need to explicitly tell Java application (which MA really is) what kind of characterset it should use to read those text properly. In order to do that just add to a script calling MA the following line:  export JAVA_TOOL_OPTIONS=-Dfile.encoding=<JAVA_ISO_ENCODING>  when the particular encoding must match the NLS_LANG in your Forms Builder environment (for example for Polish characterset you need to use ISO-8859-2).Second issue you can encounter related to national charactersets is lack of national symbols in you reports after migration. This can be solved by adding appropriate NLS_LANG entry in your reports environment. Sometimes instead of particular characterset you see "Greek characters" in your reports. This is just default font used by reports engine instead of the one defined in your report. To solve it you must copy fonts definitions from your old environment (e.g. Forms 10g installation) to appropriate directory in new installation (usually AFM folder). For more information about this and other issues please refer to https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&doctype=BULLETIN&id=1297012.1at My Oracle Support site. That's all for today, stay tuned for more posts on this topic! Lukasz

    Read the article

  • Android Touch Event Collision Detection

    - by chrissb
    I'm relatively new to both Java and Android, so hopefully the problem I'm having is stemming from something pretty minor that I've overlooked. I've got a (very early stage) game that I've started working on, for Android using Java. At this stage, when the user touches the screen, if they touched a point at which there is an enemy, the enemies health is decreased and they become immobile (for the current implementation at least). The issue that I'm having is that the touch detection doesn't always seem to work. I've got a testing sprite set up that goes to the eventX and eventY coordinates of the touch down event, and it always seems to collide with the enemy object. Yet, the enemy doesn't always register as being hit, and sometimes a hit is registered when the sprite indicates the touch coordinates were outside of the enemies bounding box. I realise that this probably doesn't mean much without any code, so here's what I've got so far. Be gentle, as this is literally my first attempt at something more than basic movement etc. First off, the MainGamePanel class registers the touch event, and informs the levelmanager class (which is what I set up to monitor/handle enemies) public boolean onTouchEvent(MotionEvent event) { if (event.getAction() == MotionEvent.ACTION_DOWN){ levelManager.handleActionDown((int)event.getX(), (int)event.getY()); targetX=event.getX(); targetY=event.getY(); } if (event.getAction() == MotionEvent.ACTION_MOVE) { //the gestures } if (event.getAction() == MotionEvent.ACTION_UP) { //touch was released } return true; } From there, in the levelmanager class the touch event is passed on to all of the enemies within a list array: public static void handleActionDown(int eventX,int eventY){ hit=false; for (enemy1 en : enemy1array){ en.handleActionDown(eventX, eventY); } } The rest of the collision code is handled within the enemies handleActionDown function: public void handleActionDown(int eventX, int eventY) { if(eventX>this.x-enemy1bitmap.getWidth() && eventX<this.x+enemy1bitmap.getWidth() && eventY>this.y-enemy1bitmap.getHeight() && eventY<this.x+enemy1bitmap.getHeight()){ takeDamage(1); levelmanager.setHit(); } } I should probably be using getWidth()/2 and getHeight()/2 for it to be more accurate, but I expanded the area to test this - although I've noticed no improvement. At this stage, the games detection over whether or not the enemy is hit is spotty at best. Generally it takes two or three attempts before a collision is successfully registered, even though the sprite that is being used for testing and set to the eventX and eventY coordinates always indicates that the collision should have worked. Hopefully someone can steer me in the right direction here, and if more information is needed, ask away! Cheers, -Chris

    Read the article

  • Determining cause of random latency and loading issues

    - by Sherwin Flight
    I'm not sure exactly what details to post in regards to my issue, because I'm not sure what is relevant. Prior to the end of September my websites all loaded quickly, in almost all cases. Loading time wasn't usually more than a few seconds. However, since the end of September I noticed a big increase in page loading times. In some cases pages were taking 30 seconds or more to load. I do have a remote monitoring service monitoring some of the sites as well, and the image below shows the response times over the past month. The response times shown at the beginning of this graph were what the usual response times were prior to this issue occurring. You can see that there has been a significant increase in response times from the beginning to the end of this graph. The thing is, the problem is not happening 100% of the time. If I click through the site, or even just keep refreshing the page, about 25% of the time the pages load quickly, the remaining 75% of the time they load slowly. Sometimes the pages take so long to load that they time out, and don't load at all. I have contacted my hosting provider, and they said things at their end was fine. I don't believe the problem is my home internet provider, because all other websites load without a problem. The server is located in Texas, USA. This also raises another interesting point. My remote monitor checks my site from two locations, California, USA, and London, England. As you can see in the chart below the response time is actually shorter when checked from London, which doesn't seem to make sense, since the server is physically closer to the California monitoring location. I would have expected the London monitoring location to have higher response times since they are physically farther away. I should also point out that in some traceroute test I've done it seem like the first connection to the server seems to take the longest, then after that the rest of the page loads quickly. Below is a little chart showing the times for the first connection to the server. So, what could be causing this problem, and what steps can I take to resolve it or at least narrow down the problem? Sending the request to the server was very quick, and receiving the reply back seems pretty quick, but the WAIT time is really long. So it connects, sends the request, but then waits close to 30 seconds before it starts receiving data back. I am also aware that there are things I can do to speed up page loading times, like reducing the number of CSS and JS files used on a page, compressing images, etc. This is not really what the source of the problem is though, because nothing has really changed on the site since before the problem started, and other sites on the same server are loading slowly as well.

    Read the article

  • How do I stop icons appearing on the desktop in a particular area?

    - by Seamus
    When I download something to my desktop, or insert a CD or flash drive, the icon appears on my desktop. When I have conky running, the icon sometimes appears in the top right corner, underneath conky; where I can't see it. How do I stop this happening? My .conkyrc is pasted below. I didn't write it all myself, so I'm not entirely sure what I need to change, or what parts are relevant for this particular question... # UBUNTU-CONKY # A comprehensive conky script, configured for use on # Ubuntu / Debian Gnome, without the need for any external scripts. # # Based on conky-jc and the default .conkyrc. # INCLUDES: # - tail of /var/log/messages # - netstat shows number of connections from your computer and application/PID making it. Kill spyware! # # -- Pengo # # Create own window instead of using desktop (required in nautilus) own_window yes own_window_type override own_window_transparent yes own_window_hints undecorated,below,sticky,skip_taskbar,skip_pager # Use double buffering (reduces flicker, may not work for everyone) double_buffer yes # fiddle with window use_spacer right # Use Xft? use_xft yes xftfont DejaVu Sans:size=8 xftalpha 0.8 text_buffer_size 2048 # Update interval in seconds update_interval 3.0 # Minimum size of text area # minimum_size 250 5 # Draw shades? draw_shades no # Text stuff draw_outline no # amplifies text if yes draw_borders no uppercase no # set to yes if you want all text to be in uppercase # Stippled borders? stippled_borders 3 # border margins border_margin 9 # border width border_width 10 # Default colors and also border colors, grey90 == #e5e5e5 default_color grey own_window_colour brown own_window_transparent yes # Text alignment, other possible values are commented #alignment top_left alignment top_right #alignment bottom_left #alignment bottom_right # Gap between borders of screen and text gap_x 10 gap_y 20 # stuff after 'TEXT' will be formatted on screen TEXT $color ${color orange}SYSTEM ${hr 2}$color $nodename $sysname $kernel on $machine ${color orange}CPU ${hr 2}$color ${freq}MHz Load: ${loadavg} Temp: ${acpitemp} $cpubar ${cpugraph 000000 ffffff} NAME ${goto 150}PID ${goto 200}CPU% ${goto 250}MEM% ${top name 1} ${goto 150}${top pid 1} ${goto 200}${top cpu 1} ${goto 250}${top mem 1} ${top name 2} ${goto 150}${top pid 2} ${goto 200}${top cpu 2} ${goto 250}${top mem 2} ${top name 3} ${goto 150}${top pid 3} ${goto 200}${top cpu 3} ${goto 250}${top mem 3} ${top name 4} ${goto 150}${top pid 4} ${goto 200}${top cpu 4} ${goto 250}${top mem 4} ${color orange}MEMORY / DISK ${hr 2}$color RAM: $memperc% ${membar 6}$color Swap: $swapperc% ${swapbar 6}$color Home: ${fs_free_perc /home}% ${fs_bar 6 /}$color Free Space: ${fs_free /home} ${color orange}NETWORK (${addr eth0}) ${hr 2}$color Down: $color${downspeed eth0} k/s ${alignr}Up: ${upspeed eth0} k/s ${downspeedgraph eth0 25,140 000000 ff0000} ${alignr}${upspeedgraph eth0 25,140 000000 00ff00}$color Total: ${totaldown eth0} ${alignr}Total: ${totalup eth0} ${execi 30 netstat -ept | grep ESTAB | awk '{print $9}' | cut -d: -f1 | sort | uniq -c | sort -nr} ${color orange}WIRELESS (${addr wlan0}) ${hr 2}$color Down: $color${downspeed wlan0} k/s ${alignr}Up: ${upspeed wlan0} k/s ${downspeedgraph wlan0 25,140 000000 ff0000} ${alignr}${upspeedgraph wlan0 25,140 000000 00ff00}$color Total: ${totaldown wlan0} ${alignr}Total: ${totalup wlan0} ${execi 30 netstat -ept | grep ESTAB | awk '{print $9}' | cut -d: -f1 | sort | uniq -c | sort -nr} Conky solutions have been offered, but perhaps these aren't the best way of approaching it. What I really want is to stop icons even appearing in that part of the desktop window: that is, I want to make part of the desktop real estate "off-limits" to new icons appearing on the desktop.

    Read the article

  • how do I get dual monitors to work properly in Ubuntu 11.10 on a Dell Latitude D630?

    - by wes cook
    I have spent a lot of time trying to get dual monitors to work on Ubuntu 11.10 on my Dell Latitude D630 (nVidia NVS 135m video card). - For starters, the System Displays settings app always only showed one unknown monitor, even though I had the external Acer monitor connected. - So I downloaded and installed the nVidia drivers. According to what I read I would need to only use the nVidia driver app (nVidia X Server Settings), so that's what I've done. (System Displays settings continued to only show a single monitor anyway). - nVidia settings app only showed on monitor until I changed the BIOS setting to use the onboard video for external monitor (not the dock video, which it was set to, even though I don't have a docking station). - The nVidia setting app now recognized both monitors. So, I setup the X Server display config as Separate X screen for both monitors. My laptop screen shows up as AUO 1440x900 and my external monitor as Acer E211H 1920x1080. - Everything seemed like it would work, but the external monitor was just a complete white screen. The external monitor was non-functional, even though sometimes it would show the background image - still nothing would show up over there. - So, I checked the Enable Xinerama box. - Now, after logging out and back in, the wallpaper extends to both screens but I get no taskbar at the bottom or top, no system menus, and I have to press the power button to restart or log off. - After experimenting with all the shells, the only one that shows the menus and taskbars when I log in is Gnome Classic. - This is pretty much the same symptoms as found here: How do I fix 11.10 GUI?. - So, I resign myself to the older shell. - Everything works fine until ... I unplug the external monitor ... this is a laptop after all. - Anyway, after doing some work on the road, I plug back in and I still see both screens and it's functional except, ... - Now, the laptop screen (with the taskbar and menu bar) has 4 black bars at the top that windows cannot cover. The top bar is the menu bar (with Applications, Places, the date and time and the system menu on the right). But the next 3 bars (the same height as the top menu bar) are empty and are just reducing the max size of windows on that screen. - See screenshot here: http://i39.tinypic.com/35d2kh1.png - So ... 1. How do I get rid of those extra 3 black bars? They're taking valuable screen space. 2. (less critical) How do I successfully use both screens in the Ubuntu or Ubuntu 2D shell?

    Read the article

  • More on PHP and Oracle 11gR2 Improvements to Client Result Caching

    - by christopher.jones
    Oracle 11.2 brought several improvements to Client Result Caching. CRC is way for the results of queries to be cached in the database client process for reuse.  In an Oracle OpenWorld presentation "Best Practices for Developing Performant Application" my colleague Luxi Chidambaran had a (non-PHP generated) graph for the Niles benchmark that shows a DB CPU reduction up to 600% and response times up to 22% faster when using CRC. Sometimes CRC is called the "Consistent Client Cache" because Oracle automatically invalidates the cache if table data is changed.  This makes it easy to use without needing application logic rewrites. There are a few simple database settings to turn on and tune CRC, so management is also easy. PHP OCI8 as a "client" of the database can use CRC.  The cache is per-process, so plan carefully before caching large data sets.  Tables that are candidates for caching are look-up tables where the network transfer cost dominates. CRC is really easy in 11.2 - I'll get to that in a moment.  It was also pretty easy in Oracle 11.1 but it needed some tiny application changes.  In PHP it was used like: $s = oci_parse($c, "select /*+ result_cache */ * from employees"); oci_execute($s, OCI_NO_AUTO_COMMIT); // Use OCI_DEFAULT in OCI8 <= 1.3 oci_fetch_all($s, $res); I blogged about this in the past.  The query had to include a specific hint that you wanted the results cached, and you needed to turn off auto committing during execution either with the OCI_DEFAULT flag or its new, better-named alias OCI_NO_AUTO_COMMIT.  The no-commit flag rule didn't seem reasonable to me because most people wouldn't be specific about the commit state for a query. Now in Oracle 11.2, DBAs can now nominate tables for caching, either with CREATE TABLE or ALTER TABLE.  That means you don't need the query hint anymore.  As well, the no-commit flag requirement has been lifted.  Your code can now look like: $s = oci_parse($c, "select * from employees"); oci_execute($s); oci_fetch_all($s, $res); Since your code probably already looks like this, your DBA can find the top queries in the database and simply tune the system by turning on CRC in the database and issuing an ALTER TABLE statement for candidate tables.  Voila. Another CRC improvement in Oracle 11.2 is that it works with DRCP connection pooling. There is some fine print about what is and isn't cached, check the Oracle manuals for details.  If you're using 11.1 or non-DRCP "dedicated servers" then make sure you use oci_pconnect() persistent connections.  Also in PHP don't bind strings in the query, although binding as SQLT_INT is OK.

    Read the article

  • PowerShell – Show a Notification Balloon

    - by BuckWoody
    In my presentations for PowerShell I sometimes want to start a process (like a backup) that will take some time. I normally pop up a notification “balloon” at the start, and then do the bulk of the work, and then pop up a balloon at the end to let me know it’s done. You can actually try out this little sample (on a test system, of course) without any other code to see what it does. Then just put the other PowerShell commands in the #Do Some Work part. Oh – throw an icon (.ico file) in a c:\temp directory or point that somewhere else. (No, this probably isn’t original. Can’t remember where I saw the original code, but I’ve modified it a bit anyway, so if you’re the original author and this looks slightly familiar, post a comment.) [void] [System.Reflection.Assembly]::LoadWithPartialName("System.Windows.Forms") $objBalloon = New-Object System.Windows.Forms.NotifyIcon $objBalloon.Icon = "C:\temp\Folder.ico" # You can use the value Info, Warning, Error $objBalloon.BalloonTipIcon = "Info" # Put what you want to say here for the Start of the process $objBalloon.BalloonTipTitle = "Begin Title" $objBalloon.BalloonTipText = "Begin Message" $objBalloon.Visible = $True $objBalloon.ShowBalloonTip(10000) # Do some work # Put what you want to say here for the completion of the process $objBalloon.BalloonTipTitle = "End Title" $objBalloon.BalloonTipText = "End Message" $objBalloon.Visible = $True $objBalloon.ShowBalloonTip(10000) Script Disclaimer, for people who need to be told this sort of thing: Never trust any script, including those that you find here, until you understand exactly what it does and how it will act on your systems. Always check the script on a test system or Virtual Machine, not a production system. Yes, there are always multiple ways to do things, and this script may not work in every situation, for everything. It’s just a script, people. All scripts on this site are performed by a professional stunt driver on a closed course. Your mileage may vary. Void where prohibited. Offer good for a limited time only. Keep out of reach of small children. Do not operate heavy machinery while using this script. If you experience blurry vision, indigestion or diarrhea during the operation of this script, see a physician immediately. Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • RIM's current BB7 developer toolset is a joke

    - by mbrit
    tl;dr - RIM's current developer toolset is not fit for purpose.Background to this is that I'm currently working on a PhoneGap/Cordova project for a client that has to run on BlackBerry. The tooling is so ridiculous to use that even though I had a gentle dig at them in a Guardian piece it's worth having a more full-on attack.At the moment, RIM's pitch is that apps are built for the current BBOS7 devices using WebWorks. This is an HTML-based toolset. Essentially a browser is spun up in a native app container and your app is powered by JavaScript. Specific JavaScript libraries exist that thunk down to native capabilities no the device. I happen to use PhoneCap/Cordova in combination with this.The tooling is non-existent. I'm using TextMate, Ant, and Terminal to develop the app. There's no "console.log" output, and no debugging. The only way to instrument the app is to put "alert" calls in your code.Apart from the fact that that's *not* fine in 2012, how about this… every time you deploy a new app to the device, the device has to reboot. This process takes six minutes on a relatively modern BlackBerry device. How about this as well - in order to get a file into the package it has to be signed. My small app over here has 100 different files (75 or so generated). Signing doesn't happen locally, it happens on RIM's servers in Waterloo. Thus whenever you deploy the app you have this utility have to call RIM's servers 100 times. More to the point, sometimes during the day these servers have "micro-downtime" moments where they're unreachable for five or ten minutes, normally two or three times a day. Oh yes, you'll also get an email sent to you per signing on success or failure. 100 inbound emails, per deployment.(I started this post at the beginning of one of these cycles, by the way. That's how long it takes to build and deploy *once*. By the way, the change I made didn't work.)To clarify:* Change the script,* Build it using Ant,* Ant will spin up a Java app that talks to RIM's servers to sign it.* Receive 100 emails, assuming the server is up.* App deployed - takes about 30 seconds.* BlackBerry device restarts - takes about six minutes.* Find and open the app. Go through security prompts.* Test the app, with no "console.log" output and no debugger."Why not use the simulator?" I hear you ask. Well, apart from the fact that the simulator refused to reach any network service over HTTPS that I happen to own? (Some people suggest changing DNS settings for this known issue.) Admittedly, the simulator does show you console.log, but you still have the "six minute" restart issue on the simulator.Developers will understand this problem. Breaking concentration for six-plus minutes every time you want to deploy an app turns developing into a nightmare. Combining that with no worthy debugging tools turns the toolset into a joke.

    Read the article

  • Send an E-mail Quickly with the GmailThis! Bookmarklet

    - by Asian Angel
    Sometimes you need to send out a quick e-mail for a project that you are working on, something really important that you just remembered, or perhaps just a note to yourself. If you use Gmail and like keeping things simple then join us as we look at the GmailThis! Bookmarklet. GmailThis! in Action To get set up all that you need to do is visit the webpage (link provided below) and drag the bookmarklet to your “Bookmarks Toolbar”. For our example we decided to go with the “personal note” approach. As you can see here we selected/highlighted a portion of the text and then clicked on our new bookmarklet. The bookmarklet will automatically copy and paste the name of the webpage, the URL, and any text that you selected/highlighted into the new e-mail. A nice feature that we liked was that it opened in a new temporary window to help focus on composing our letter. This is what you will see when you have finished your letter and clicked “Send”. The window will automatically close itself after a few seconds so that you do not even have to worry with it afterwards. Looking at our “Inbox” there is our new e-mail looking oh so nice. Conclusion If you need to send out a quick e-mail using your Gmail account then this bookmarklet makes it as quick and simple as possible. This is definitely one to add to your bookmarklets collection. Links Get the GmailThis! Bookmarklet for your browser Similar Articles Productive Geek Tips How to Send and Receive Hotmail from Your Gmail AccountShare Your Favorite Webpages with the AddThis BookmarkletPower Up and Manage Your Windows Send To Menu with Send To ToysTurn off New Mail Notification for PocoMail Junk Mail FolderCreate an Email Template in Outlook 2003 TouchFreeze Alternative in AutoHotkey The Icy Undertow Desktop Windows Home Server – Backup to LAN The Clear & Clean Desktop Use This Bookmarklet to Easily Get Albums Use AutoHotkey to Assign a Hotkey to a Specific Window Latest Software Reviews Tinyhacker Random Tips DVDFab 6 Revo Uninstaller Pro Registry Mechanic 9 for Windows PC Tools Internet Security Suite 2010 Monitor Applications With Kiwi LocPDF is a Visual PDF Search Tool Download Free iPad Wallpapers at iPad Decor Get Your Delicious Bookmarks In Firefox’s Awesome Bar Manage Photos Across Different Social Sites With Dropico Test Drive Windows 7 Online

    Read the article

  • Oracle went back to school !....

    - by Cristina Ciocoiu
    I am Georgiana, Contracts Manager for Oracle University and Advanced Customer Services in Romania. I started working for Oracle for 4 years ago as a Contracts Specialist. Two years ago I became a manager of a team of 9 Contracts Specialists. On a sunny day in March some members of my team visited the students of the Academy of Economic Studies, accompanied by Recruitment colleagues. This was part of a new initiative to raise awareness on career opportunities at Oracle. We spent approximately 2 hours illustrating and explaining different aspects of the day-to-day activities of an Oracle Contracts Specialist to the future graduates of the Academy. Role Play Since a role play is worth 1000 job descriptions, the audience witnessed an entertaining performance on the contracting process from the phase of the negotiation with the customer to actual signing of the contract. The main focus was on the role of Contracts Specialist liaising with all the groups involved and ensuring that the contract is compliant with Oracle policies while generating the expected revenue. However, the team took other roles as well i.e. Sales Representative, Customer, Business Approver and Lawyer to demonstrate their role in the process. As each of these roles only have a small slice of the big pie, it is vital to understand what happens before and after you come on stage as a Contract Specialist. Contracts Specialist Being a Contracts Specialist goes beyond simply knowing what policies apply, it means understanding Oracle’s core business model, understanding customers’ requests and addressing them in the most effective way. The job also involves connecting smaller teams that are often geographically dispersed across multiple regions so that they become a bigger, stronger and successful team. You are the expert in this key position that can facilitate the closing of a deal or stop it from happening if the risk is too high. The role play provided insights on both. Why I love this job Events of this kind are sometimes just as useful for the “recruiters” as for the “recruits”. For me, as a presenter, it was an excellent opportunity to think about the many reasons why I love what I do in the Contracts department every day and to share this with the students. I wanted to explain to the audience, who are still considering education and career possibilities, that what we do in Contracts DOES make a difference. You have the power to achieve targets that you did not think reachable before. Working in the dynamic Oracle environment shapes you as a person and there is a lot to take away from this experience. Looking back to my years in the Academy (I graduated from the Academy myself), I wish I could have listened to more people talking about their great jobs and about how I could get there. If those were Oracle people I might have been writing this article sooner. J If you are interested to join the Contracts team please click here for more information or contact lavinia.protopopescu-AT-oracle-DOT-com. You can find all openings in Romania via http://campus.oracle.com

    Read the article

  • Using multiple diagrams per model in Entity Framework 5.0

    - by nikolaosk
    I have downloaded .Net framework 4.5 and Visual Studio 2012 since it was released to MSDN subscribers on the 15th of August.For people that do not know about that yet please have a look at Jason Zander's excellent blog post .Since then I have been investigating the many new features that have been introduced in this release.In this post I will be looking into theIn order to follow along this post you must have Visual Studio 2012 and .Net Framework 4.5 installed in your machine.Download and install VS 20120 using this link.My machine runs on Windows 8 and Visual Studio 2012 works just fine. I have also installed in my machine SQL Server 2012 developer edition. I have also downloaded and installed AdventureWorksLT2012 database.You can download this database from the codeplex website.   Before I start showcasing the demo I want to say that I strongly believe that Entity Framework is maturing really fast and now at version 5.0 can be used as your data access layer in all your .Net projects.I have posted extensively about Entity Framework in my blog.Please find all the EF related posts here. In this demo I will show you how to split an entity model into multiple diagrams using the new enhanced EF designer. We will not build an application in this demo.Sometimes our model can become too large to edit or view.In earlier versions we could only have one diagram per EDMX file.In EF 5.0 we can split the model into more diagrams.1) Launch VS 2012. Express edition will work fine.2) Create a New Project. From the available templates choose a Web Forms application  3) Add a new item in your project, an ADO.Net Entity Data Model. I have named it AdventureWorksLT.edmx.Then we will create the model from the database and click Next.Create a new connection by specifying the SQL Server instance and the database name and click OK.Then click Next in the wizard.In the next screen of the wizard select all the tables from the database and hit Finish.4) It will take a while for our .edmx diagram to be created. When I select an Entity (e.g Customer) from my diagram and right click on it,a new option appears "Move to new Diagram".Make sure you have the Model Browser window open.Have a look at the picture below 5) When we do that a new diagram is created and our new Entity is moved there.Have a look at the picture below  6) We can also right-click and include the related entities. Have a look at the picture below. 7) When we do that the related entities are copied to the new diagram.Have a look at the picture below  8) Now we can cut (CTRL+X) the entities from Diagram2 and paste them back to Diagram1.9) Finally another great enhancement of the EF 5.0 designer is that you can change colors in the various entities that make up the model.Select the entities you want to change color, then in the Properties window choose the color of your choice. Have a look at the picture below. To recap we have demonstrated how to split your entity model in multiple diagrams which comes handy in EF models that have a large number of entities in them Hope it helps!!!!

    Read the article

  • A strong component keeps everything together

    - by Justin Paul-Oracle
    Most of the times you implement a WebCenter Content based system, you require some sort of customization. Sometimes these customizations need a Java class or two, or libraries (for example, the JavaMail API), or Database Objects (like new tables, views, indexes, etc). I have seen that libraries and Database Objects are usually put in place using manual steps. This means that the library jar files are copied to one of the common classes directory (set in the Content CLASSPATH variable) and/or the database scripts are executed manually. I have also seen people place the custom Java classes in the common classes directory. While this may seem like an easy solution, think about a scenario where you need to disable or uninstall the component or if you have to upgrade or migrate the system. You have to keep these manual steps documented and execute them every time you encounter the above scenarios. It is very common that some of these manual steps are missed when you have multiple teams and people working on the system. Here are a few points to ponder upon: Place all your custom Java classes within your component. Create a new directory, say ${COMPONENT_DIR}/classes, and place your code there. You can choose to bundle all your classes into a jar or you can place the entire class directory structure. Add a path entry to the Build Settings so that it is bundled with the component when you build it. You also need to update the Custom Class Path and the Custom Class Path Load Order under the Advanced Build Settings. This will ensure that the system CLASSPATH is updated to add this new directory. Create a new component for any new library that you want to add. Add the appropriate path entries to the Build Settings so that it is bundled with the component when you build it. You also need to update the Custom Class Path, Custom Class Path Load Order and/or the Custom Library Path under the Advanced Build Settings. Enter a comma separated list of features that this component will provide. When you create other components that will use the features exposed by this component, make sure that you specify a dependency to this library component by specifying the comma separated list of features in the Advanced Build Settings. The component wizard allows you to create custom install/uninstall Java code. The wizard will create a install filter class when you check the “Has Install” checkbox on the “Install/Uninstall Settings” tab. Consider using this filter class to create database objects when you install the component and drop the objects when you uninstall the component. If you do a lot of custom component development, consider creating a install/uninstall Java class, which can execute queries defined within the component. To sum up, whenever you write a new custom component, make sure that you bundle everything within the component.

    Read the article

  • What if you could work on anything you wanted?

    - by Nick Harrison
    What if you could work on anything you wanted? Redgate is doing an experiment of sorts this week.  Called Down Tools Week.    The idea is that they stopped working on their regular projects for a week and strike out on something that catches their attention and drives their passion. Evidently in many cases, these projects have turned out to be new features in their existing products that individual were interested in, some were internal iniatives and some where evidently off the wall new ideas.   Today is show and tell where they will share with each other what they have been working on. There may well be some interesting announcements coming out of this.    The prospects are exciting. I understand that Google does something similar allowing their employees a specified amount of time to work on projects of their own choosing.    This has been the breeding ground for some of my favorite services. It is a shame that more companies do not follow such practices.   Now I know that most companies cannot afford to shut down everything for a week and sometimes you can't really explore an interesting idea in 8 hours a week or however much time Google allocates, but still it may be worth while. What would happen if your company gave you as an individual 1 week each quarter to work on a project of your own design and see what happens?   I would be happen if you still had to get approval for before your week long adventure. Personally, I think that this could be a very effective use of training budgets.   Give me a week to research something on my own and you would be amazed at what I can find out.    Maybe this should be the prerequisite before starting a new project.   Stagger the team onboarding but have everyone spend a week long sabbatical studying BizTalk before starting a project that will hinge on BizTalk. The show and tell afterwards is a great way to keep everyone honest or at least reassure management that everyone is honest.    If your goal was to spend a week researching and exploring a new technology and you had to do a show and tell afterwards to show off what you had learned, then everyone can learn a bit of what you just learned.     Sounds like a promising win win for me. Maybe it is a pipe dream, but what if .... What would you work on if given the opportunity to work on anything you wanted?

    Read the article

  • Walmart's Mobile Self-Checkout

    - by David Dorf
    Reuters recently reported that Walmart was testing an iPhone-based self-checkout at a store near its headquarters.  Consumers scan items as they're placed in the physical basket, then the virtual basket is transferred to an existing self-checkout station where payment is tendered.  A very solid solution, but not exactly original. Before we go further, let's look at the possible cost savings for Walmart.  According to the article: Pushing more shoppers to scan their own items and make payments without the help of a cashier could save Wal-Mart millions of dollars, Chief Financial Officer Charles Holley said on March 7. The company spends about $12 million in cashier wages every second at its Walmart U.S. stores. Um, yeah. Using back-of-the-napkin math, I calculated Walmart's cashiers are making $157k per hour.  A more accurate statement would be saving $12M per year for each second saved on the average transaction time.  So if this self-checkout approach saves 2 seconds per transaction on average, Walmart would save $24M per year on labor.  Maybe.  Sometimes that savings will be used to do other tasks in the store, so it may not directly translate to less employees. When I saw this approach demonstrated in Sweden, there were a few differences, which may or may not be in Walmart's plans.  First, the consumers were identified based on their loyalty card.  In order to offset the inevitable shrink, retailers need to save on labor but also increase basket size, typically via in-aisle promotions.  As they scan items, retailers should target promos, and that's easier to do if you know some shopping history.  Last I checked, Walmart had no loyalty program. Second, at the self-checkout station consumers were randomly selected for an audit in which they must re-scan all the items just like you do at a typical self-checkout.  If you were found to be stealing, your ability to use the system can be revoked.  That's a tough one in the US, especially when the system goes wrong, either by mistake or by lying.  At least in my view, the Swedes are bit more trustworthy than the people of Walmart. So while I think the idea of mobile self-checkout has merit, perhaps its not right for Walmart.

    Read the article

  • Guiding Management to the Correct Decision

    - by Blumer
    My supervisor (also a developer) and I have a running joke about writing a book called "Managing From Beneath: Subversively Guiding Management to the Right Decision" and including a number of "techniques" we've developed for helping those who make the decisions to make the right ones. So far, we've got (cynicism warning!): BIC It! BIC stands for "Bury In Committee." When a bad idea comes up that someone wants to champion, we try to get it deferred to a committee for input. Typically it will either get killed outright (especially if other members of the committee are competing for you as a resource), or it will be hung up long enough that the proponent forgets about it. Smart, Stupid, or Expensive? When someone gets a visionary idea, offer them three ways to do it: a smart way, a stupid way, and an expensive way. The hope is that you've at least got a 2/3 shot of not having to do it the way that makes a piece of your soul die. All-Pro. It's a preemptive pro/con list in which you get into the mind of the (pr)opponent and think what would be cons against doing it your way. Twist them into pros and present them in your pro list before they have a chance to present them as cons. Dependicitis. Link pending decisions together, ideally with the proponent's pet project as the final link in the chain. Use this leverage to force action on those that have been put off. Preemptive Acceptance. Sometimes it's clear that management is going to go a particular direction regardless of advice to the contrary, and it's time to make the best of it. Take the opportunity to get something else you need, though. Approach the sponsor out of the blue and take the first step: "You know, I've been thinking about it, and while it's not the route I would advise, as long as we can get the schedule and budget for Project Awesome loosened up, I can work some magic to make your project fly." So ... what techniques have you come up with to try to head off the problem projects or make the best of what may come?

    Read the article

  • Parallel MSBuild FTW - Build faster in parallel

    - by deadlydog
    Hey everyone, I just discovered this great post yesterday that shows how to have msbuild build projects in parallel Basically all you need to do is pass the switches “/m:[NumOfCPUsToUse] /p:BuildInParallel=true” into MSBuild. Example to use 4 cores/processes (If you just pass in “/m” it will use all CPU cores): MSBuild /m:4 /p:BuildInParallel=true "C:\dev\Client.sln" Obviously this trick will only be useful on PCs with multi-core CPUs (which we should all have by now) and solutions with multiple projects; So there’s no point using it for solutions that only contain one project.  Also, testing shows that using multiple processes does not speed up Team Foundation Database deployments either in case you’re curious Also, I found that if I didn’t explicitly use “/p:BuildInParallel=true” I would get many build errors (even though the MSDN documentation says that it is true by default). The poster boasts compile time improvements up to 59%, but the performance boost you see will vary depending on the solution and its project dependencies.  I tested with building a solution at my office, and here are my results (runs are in seconds): # of Processes 1st Run 2nd Run 3rd Run Avg Performance 1 192 195 200 195.67 100% 2 155 156 156 155.67 79.56% 4 146 149 146 147.00 75.13% 8 136 136 138 136.67 69.85%   So I updated all of our build scripts to build using 2 cores (~20% speed boost), since that gives us the biggest bang for our buck on our solution without bogging down a machine, and developers may sometimes compile more than 1 solution at a time.  I’ve put the any-PC-safe batch script code at the bottom of this post. The poster also has a follow-up post showing how to add a button and keyboard shortcut to the Visual Studio IDE to have VS build in parallel as well (so you don’t have to use a build script); if you do this make sure you use the .Net 4.0 MSBuild, not the 3.5 one that he shows in the screenshot.  While this did work for me, I found it left an MSBuild.exe process always hanging around afterwards for some reason, so watch out (batch file doesn’t have this problem though).  Also, you do get build output, but it may not be the same that you’re used to, and it doesn’t say “Build succeeded” in the status bar when completed, so I chose to not make this my default Visual Studio build option, but you may still want to. Happy building! ------------------------------------------------------------------------------------- :: Calculate how many Processes to use to do the build. SET NumberOfProcessesToUseForBuild=1  SET BuildInParallel=false if %NUMBER_OF_PROCESSORS% GTR 2 (                 SET NumberOfProcessesToUseForBuild=2                 SET BuildInParallel=true ) MSBuild /maxcpucount:%NumberOfProcessesToUseForBuild% /p:BuildInParallel=%BuildInParallel% "C:\dev\Client.sln"

    Read the article

  • Phone number in meta description bad or good for local rankings NAP

    - by bybe
    Once again I'm at it with increasing people's local rankings and I've learnt so much about local rankings in the past 2 weeks it feels like my brain is gonna pop anyway, question is fairly simple for someone who engages in local rankings and I appreciate the question may be a little guess work but isn't SEO mostly guessing anyway? From what I've read and learned that Google works of a system called nap for local rankings (With many other factors but this question is purely based on NAP). For people who care about local rankings NAP stands for Name of Business / Address of Business / Phone Number for Business. Now what what I've read you don't need the whole NAP to be on one website, a P or just a N can help towards your local rankings. It's believed that NAP rewards more than just P and N for example but knowing Google they might have a diversity checker which is my concern what your get to in a moment. Now of course sites weight differently where your business is posted, it's certainly going to be more credible if your NAP details are on your national phone book than say a blog site, so taking in this consideration too. Pure Guess (Not apart of the question but none the less makes a good read on my belief). Now my guess work would make me believe that the formula would look something like (N)+(A)+(P)x(T) So (N)name would be 1 or 0 to indicate present or not So (A)dress would be 1 or 0 to indicate present or not So (P)hone would be 1 or 0 to indicate present or not So (T)rust would be 1-100 to indicate level of trust So a phone number appearing on youtube might look something like 0+0+1x95= 95 and a NAP appearing on your national phone book might look something like 1+1+1x100= 300 Please note that I'm not saying this is the sole factor and I'm sure its way more complex that this with things like other factors on the page, off the page (Reviews, Links, Clicks) and so on but its still a contributor). The Question My question is fairly simple and I'd imagine hard to impossible to have an actual definite answer to this but maybe someone has seen official wording else where on this, is it bad to include address or phone number in the Meta Description? The reason I ask is that one of my competitors has these elements in the meta descriptions and their local rankings are absolutely superb, the problem I have with this is scrap bot sites like 'Similar Too' 'Seo Rankings' and 1,000's of the other scrap box networks that scrap site and then make urls with your site information are mostly limited to your meta description what this means that your phone number, address and sometimes even your company name if the domain is exact will appear as AP, and even NAP on thousands of websites. So, is it a bad strategy to include phone number and address in meta description, everything I read into would suggest its good of course with the downside of maybe lowering quality of description for click thoughts but top rankings would increase this 10 folds anyhow..

    Read the article

  • Learn programming backwards, or "so I failed the FizzBuzz test. Now what?"

    - by moraleida
    A Little Background I'm 28 today, and I've never had any formal training in software development, but I do have two higher education degrees equivalent to a B.A in Public Relations and an Executive MBA focused on Project Management. I've worked on those fields for about 6 years total an then, 2,5 years ago I quit/lost my job and decided to shift directions. After a month thinking things through I decided to start freelancing developing small websites in WordPress. I self-learned my way into it and today I can say I run a humble but successful career developing themes and plugins from scratch for my clients - mostly agencies outsourcing some of their dev work for medium/large websites. But sometimes I just feel that not having studied enough math, or not having a formal understanding of things really holds me behind when I have to compete or work with more experienced developers. I'm constantly looking for ways to learn more but I seem to lack the basics. Unfortunately, spending 4 more years in Computer Science is not an option right now, so I'm trying to learn all I can from books and online resources. This method is never going to have NASA employ me but I really don't care right now. My goal is to first pass the bar and to be able to call myself a real programmer. I'm currently spending my spare time studying Java For Programmers (to get a hold on a language everyone says is difficult/demanding), reading excerpts of Code Complete (to get hold of best practices) and also Code: The Hidden Language of Computer Hardware and Software (to grasp the inner workings of computers). TL;DR So, my current situation is this: I'm basically capable of writing any complete system in PHP (with the help of Google and a few books), integrating Ajax, SQL and whatnot, and maybe a little slower than an experienced dev would expect due to all the research involved. But I was stranded yesterday trying to figure out (not Google) a solution for the FizzBuzz test because I didn't have the if($n1 % $n2 == 0) method modulus operator memorized. What would you suggest as a good way to solve this dilemma? What subjects/books should I study that would get me solving problems faster and maybe more "in a programmers way"? EDIT - Seems that there was some confusion about what did I not know to solve FizzBuzz. Maybe I didn't express myself right: I knew the steps needed to solve the problem. What I didn't memorize was the modulus operator. The problem was in transposing basic math to the program, not in knowing basic math. I took the test for fun, after reading about it on Coding Horror. I just decided it was a good base-comparison line between me and formally-trained devs. I just used this as an example of how not having dealt with math in a computer environment before makes me lose time looking up basic things like modulus operators to be able to solve simple problems.

    Read the article

  • What do you do when practical problems get in the way of practical goals?

    - by P.Brian.Mackey
    UPDATE Source control is good to use. Sometimes, real world issues make it impractical to use. For example: If the team is not used to using source control, training problems can arise If a team member directly modifies code on the server, various issues can arise. Merge problems, lack of history, etc Let's say there's a project that is way out of sync. The physical files on the server differ in unknown ways over ~100 files. Merging would take not only a great knowledge of the project, but is also well beyond the ability to complete in the given time. Other projects are falling out of sync. Developers continue to have a distrust of source control and therefore compound the issue by not using source control. Developers argue that using source control is wasteful because merging is error prone and difficult. This is a difficult point to argue, because when source control is being so badly mis-used and source control continually bypassed, it is error prone indeed. Therefore, the evidence "speaks for itself" in their view. Developers argue that directly modifying source control saves time. This is also difficult to argue. Because the merge required to synchronize the code to start with is time consuming, across ~10 projects. Permanent files are often stored in the same directory as the web project. So publishing (full publish) erases these files that are not in source control. This also drives distrust for source control. Because "publishing breaks the project". Fixing this (moving stored files out of the solution subfolders) takes a great deal of time and debugging as these locations are not set in web.config and often exist across multiple code points. So, the culture persists itself. Bad practice begets more bad practice. Bad solutions drive new hacks to "fix" much deeper, much more time consuming problems. Servers, hard drive space are extremly difficult to come by. Yet, user expectations are rising. What can be done in this situation?

    Read the article

  • Make the Taskbar Buttons Switch to the Last Active Window in Windows 7

    - by The Geek
    The new Windows 7 taskbar’s Aero Peek feature, with the live thumbnails of every window, is awesome… but sometimes you just want to be able to click the taskbar button and have the last open window show up instead. Here’s a quick hack to make it work better. To better understand the problem, imagine having nine windows of the same type open on your screen, but you are primarily working in just one of the windows at a time. So every time you want to switch back, you have to click the taskbar button, and then choose the one you are using from the list, which can be pretty annoying… Now if you know your Windows 7 shortcuts, you’d know that you can simply hold down the Ctrl key while clicking on the taskbar button, and the last window will show up. In fact, you can keep holding down the Ctrl key and keep clicking, and Windows will cycle through the open windows. It’s a useful shortcut, but hardly something you want to do every single time. Instead, we’ll use a quick registry hack to make the normal click switch to the last open window—if you still want to see the thumbnail list, just hover your mouse over the button for half a second to see the full list. Manual Registry Hack for Last Active Window Open up regedit.exe through the start menu search or run box, and then head down to the following registry key: HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\Advanced Once you’re there, create a new 32-bit DWORD value on the right hand side, give it the name LastActiveClick, and set the value to 1. Once you are done, it should look something like this: Once you are done, you’ll have to log off and back on, or you can kill Explorer.exe through Task Manager and re-open it. Download the Registry Hack Instead Since you probably don’t feel like registry hacking, we’ve provided you an easy downloadable version. You can simply download the file, extract it, and then double-click on the LastActiveClick.reg file. Once you are done, you’ll have to log off and back on, just like with the manual registry hack. Download LastActiveClick Registry Hack from howtogeek.com Similar Articles Productive Geek Tips Make the Windows 7 Taskbar Work More Like Windows XP or VistaStupid Geek Tricks: Select Multiple Windows on the TaskbarReorganize Your Taskbar Buttons and Tray Icons in XP/VistaKeyboard Ninja: Create a Hotkey to Switch to Your Open Outlook WindowTaskbar Eliminator Does What the Name Implies: Hides Your Windows Taskbar TouchFreeze Alternative in AutoHotkey The Icy Undertow Desktop Windows Home Server – Backup to LAN The Clear & Clean Desktop Use This Bookmarklet to Easily Get Albums Use AutoHotkey to Assign a Hotkey to a Specific Window Latest Software Reviews Tinyhacker Random Tips DVDFab 6 Revo Uninstaller Pro Registry Mechanic 9 for Windows PC Tools Internet Security Suite 2010 Gadfly is a cool Twitter/Silverlight app Enable DreamScene in Windows 7 Microsoft’s “How Do I ?” Videos Home Networks – How do they look like & the problems they cause Check Your IMAP Mail Offline In Thunderbird Follow Finder Finds You Twitter Users To Follow

    Read the article

  • How do I stop icons appearing on the desktop under conky?

    - by Seamus
    When I download something to my desktop, or insert a CD or flash drive, the icon appears on my desktop. When I have conky running, the icon sometimes appears in the top right corner, underneath conky; where I can't see it. How do I stop this happening? My .conkyrc is pasted below. I didn't write it all myself, so I'm not entirely sure what I need to change, or what parts are relevant for this particular question... # UBUNTU-CONKY # A comprehensive conky script, configured for use on # Ubuntu / Debian Gnome, without the need for any external scripts. # # Based on conky-jc and the default .conkyrc. # INCLUDES: # - tail of /var/log/messages # - netstat shows number of connections from your computer and application/PID making it. Kill spyware! # # -- Pengo # # Create own window instead of using desktop (required in nautilus) own_window yes own_window_type override own_window_transparent yes own_window_hints undecorated,below,sticky,skip_taskbar,skip_pager # Use double buffering (reduces flicker, may not work for everyone) double_buffer yes # fiddle with window use_spacer right # Use Xft? use_xft yes xftfont DejaVu Sans:size=8 xftalpha 0.8 text_buffer_size 2048 # Update interval in seconds update_interval 3.0 # Minimum size of text area # minimum_size 250 5 # Draw shades? draw_shades no # Text stuff draw_outline no # amplifies text if yes draw_borders no uppercase no # set to yes if you want all text to be in uppercase # Stippled borders? stippled_borders 3 # border margins border_margin 9 # border width border_width 10 # Default colors and also border colors, grey90 == #e5e5e5 default_color grey own_window_colour brown own_window_transparent yes # Text alignment, other possible values are commented #alignment top_left alignment top_right #alignment bottom_left #alignment bottom_right # Gap between borders of screen and text gap_x 10 gap_y 20 # stuff after 'TEXT' will be formatted on screen TEXT $color ${color orange}SYSTEM ${hr 2}$color $nodename $sysname $kernel on $machine ${color orange}CPU ${hr 2}$color ${freq}MHz Load: ${loadavg} Temp: ${acpitemp} $cpubar ${cpugraph 000000 ffffff} NAME ${goto 150}PID ${goto 200}CPU% ${goto 250}MEM% ${top name 1} ${goto 150}${top pid 1} ${goto 200}${top cpu 1} ${goto 250}${top mem 1} ${top name 2} ${goto 150}${top pid 2} ${goto 200}${top cpu 2} ${goto 250}${top mem 2} ${top name 3} ${goto 150}${top pid 3} ${goto 200}${top cpu 3} ${goto 250}${top mem 3} ${top name 4} ${goto 150}${top pid 4} ${goto 200}${top cpu 4} ${goto 250}${top mem 4} ${color orange}MEMORY / DISK ${hr 2}$color RAM: $memperc% ${membar 6}$color Swap: $swapperc% ${swapbar 6}$color Home: ${fs_free_perc /home}% ${fs_bar 6 /}$color Free Space: ${fs_free /home} ${color orange}NETWORK (${addr eth0}) ${hr 2}$color Down: $color${downspeed eth0} k/s ${alignr}Up: ${upspeed eth0} k/s ${downspeedgraph eth0 25,140 000000 ff0000} ${alignr}${upspeedgraph eth0 25,140 000000 00ff00}$color Total: ${totaldown eth0} ${alignr}Total: ${totalup eth0} ${execi 30 netstat -ept | grep ESTAB | awk '{print $9}' | cut -d: -f1 | sort | uniq -c | sort -nr} ${color orange}WIRELESS (${addr wlan0}) ${hr 2}$color Down: $color${downspeed wlan0} k/s ${alignr}Up: ${upspeed wlan0} k/s ${downspeedgraph wlan0 25,140 000000 ff0000} ${alignr}${upspeedgraph wlan0 25,140 000000 00ff00}$color Total: ${totaldown wlan0} ${alignr}Total: ${totalup wlan0} ${execi 30 netstat -ept | grep ESTAB | awk '{print $9}' | cut -d: -f1 | sort | uniq -c | sort -nr}

    Read the article

  • Becoming the well-integrated content company (and combating AIUTLVFS)

    - by Lance Shaw
    Every single day, each of us create more and more content. Sometimes it is brand new material and many times it is iterations of existing content, but no one would argue that information and content growth is growing at an almost exponential rate. With all this content being created and stored, a number of problems naturally arise. One of the most common issues that users run into is "Am I Using The Latest Version of this File Syndrome", or AIUTLVFS. This insidious syndrome is all too common and results in ineffective, poor or downright wrong business decisions being made.  When content or files are unavailable or incorrect within the scope of key business processes, the chance for erroneous and costly business decisions is magnified even further. For many companies, the ideal scenario is to be able to connect multiple business systems, both old and new, into one common content repository.  Not only does this reduce content duplication, it also helps guarantee that everyone in various departments is working off the proverbial "same page".  Sounds simple - but for many organizations, the proliferation of file shares, SharePoint sites, and other storage silos of content keep the dream of a more efficient business a distant one. We've created some online assets to help you in your evaluation and eventual improvement of your current content management and delivery systems. Take a few minutes to check out our Online Assessment Tool.  It's quick, easy and just might provide you with insights into how you can improve your current content ecosystem. While you are there, check out our new Infographic that outlines common issues faced by companies today. Feel free to save our informative Infographic PDF and share it with business colleagues and your management to help them understand the business costs and impact of inaction. Together we can stop AIUTLVFS in its tracks and run our businesses more effectively than ever. Additionally, we hope you will take a few minutes to visit our new and informative webpages dedicated to the value of a well connected, fully integrated content management system. It's a great place to learn more about how integrating WebCenter Content into your infrastructure can lower your operational costs while boosting process and worker efficiency.

    Read the article

  • Wrong statistics in AUX_STATS$ might puzzle the optimizer

    - by Mike Dietrich
    We do recommend the creation of System Statistics for quite a long time. Since Oracle 9i the optimizer works with a CPU and IO cost based model. And in order to give the optimizer some knowledge about the IO subsystem's performance and throughput - once System Statistics are collected - they'll get stored in AUX_STATS$. For this purpose in the old Oracle 9i days some default values had been defined - and you'll still find those defaults in Oracle Database 11g Release 2 in AUX_STATS$. But these old values don't reflect the performance of modern IO systems. So it might be a good best practice post upgrade to create fresh System Statistics if you haven't done this before.  You can collect System Statistics with: exec DBMS_STATS.GATHER_SYSTEM_STATS('start'); and end it later by executing: exec DBMS_STATS.GATHER_SYSTEM_STATS('stop'); You could also run DBMS_STATS.GATHER_SYSTEM_STATS('interval', interval=>N) instead where N is the number of minutes when statistics gathering is stopped automatically. Please make sure you'll do this on a real workload period. It won't make sense to gather these values while the database is in an idle state. You should do this ideally for several hours. It doesn't affect performance in a negative way as the values are anyway collected in V$SYSSTAT and V$SESSTAT. And in case you'd like to delete the stats and revert to the old default values you'd simply execute:exec DBMS_STATS.DELETE_SYSTEM_STATS; The tricky thing in Oracle Database 11.2 - and that's why I'm actually writing this blog post today - is bug9842771. This leads to wrong values in AUX_STATS$ for SREADTIM and MREADTIM by factor 1000 guiding the optimizer sometimes into the totally wrong directon. The workaround is to overwrite these values manually and divide them by 1000. Use the DBMS_STATS.SET_SYSTEM_STATS procedure. See this MOS Note:9842771.8 for the above bug for some further information. This issue is fixed in Oracle Database 11.2.0.3 and above. To get some background information about the statistics collected in please read this section in the Oracle Database 11.2 Performance Tuning Guide. And gathering System Statistics might have some implication if you have mixed workloads - and interacts with DB_FILE_MULTIBLOCK_READ_COUNT. For more information please read section 13.4.1.2.

    Read the article

  • Pay in the future should make you think in the present

    - by BuckWoody
    Distributed Computing - and more importantly “-as-a-Service” models of computing have a different cost model. This is something that sounds obvious on the surface but it’s often forgotten during the design and coding phase of a project. In on-premises computing, we’re used to purchasing a server and all of the hardware infrastructure and software licenses needed not only for one project, but several. This is an up-front or “sunk” cost that we consume by running code the organization needs to perform its function. Using a direct connection over wires you’ve already paid for, we don’t often have to think about bandwidth, hits on the data store or the amount of compute we use - we just know more is better. In a pay-as-you-go model, however, each of these architecture decisions has a potential cost impact. The amount of data you store, the number of times you access it, and the amount you send back all come with a charge. The offset is that you don’t buy anything at all up-front, so that sunk cost is freed up. And financial professionals know that money now is worth more than money later. Saving that up-front cost allows you to invest it in other things. It’s not just that you’re using things that now cost money - it’s that the design itself in distributed computing has a cost impact. That can be a really good thing, such as when you dynamically add capacity for paying customers. If you can tie back the cost of a series of clicks to what a user will pay to do so, you can set a profit margin that is easy to track. Here’s a case in point: Assume you are using a large instance in Windows Azure to compute some data that you retrieve from a SQL Azure database. If you don’t monitor the path of the application, you may not know what you are really using. Since you’re paying by the size of the instance, it’s best to maximize it all the time. Recently I evaluated just this situation, and found that downsizing the instance and adding another one where needed, adding a caching function to the application, moving part of the data into Windows Azure tables not only increased the speed of the application, but reduced the cost and more closely tied the cost to the profit. The key is this: from the very outset - the design - make sure you include metrics to measure for the cost/performance (sometimes these are the same) for your application. Windows Azure opens up awesome new ways of doing things, so make sure you study distributed systems architecture before you try and force in the application design you have on premises into your new application structure.

    Read the article

< Previous Page | 228 229 230 231 232 233 234 235 236 237 238 239  | Next Page >