Search Results

Search found 2307 results on 93 pages for 'wild man'.

Page 59/93 | < Previous Page | 55 56 57 58 59 60 61 62 63 64 65 66  | Next Page >

  • Cat all files in a directory, with a specific file at the beginning an end...?

    - by Aeisor
    Is there a way to cat all files in a given directory, but with a particular file at the beginning and end? For example, say I have: file1.js, file2.js, file3.js, file4.js, file5.js -- Effectively I would like to cat file2.js file*.js file3.js > /var/www/output.js I've tried a few variations of these find ! -name "file2.js" ! -name "file3.js" -type f -exec cat file2.js {} file3.js > /var/www/js/output.js \; find ! -name "file2.js" ! -name "file3.js" -type f | xargs -I files cat file2.js files file3.js > /var/www/output.js but the best I can get out of it is file2.js added before and file3.js added after all other files (multiple times) I know I could specify the files in the order I wanted manually, but this is not maintainable (I'm expecting, potentially 100 files). I have looked through man cat, as well as a handful of websites devoted to xargs, find and cat to no avail. Thanks in advance.

    Read the article

  • Packaging MATLAB (or, more generally, a large binary, proprietary piece of software)

    - by nfirvine
    I'm trying to package MATLAB for internal distribution, but this could apply to any piece of software with the same architecture. In fact, I'm packaging multiple releases of MATLAB to be installed concurrently. Key things Very large installation size (~4 GB) Composed of a core, and several plugins (toolboxes) Initially, I created a single "source" package (matlab2011b) that builds several .debs (mainly matlab2011b-core and matlab2011b-toolbox-* for each toolbox). The control file is just the standard all: dh $@ There is no Makefile; only copying files. I use a number of debian/*.install files to specify files to copy from a copy of an installation to /usr/lib/. The problem is, every time I build the thing (say, to make a correction to the core package), it recopies every file listed in the *.install file to e.g debian/$packagename/usr/ (the build phase), and then has to bundle that into a .deb file. It takes a long time, on the order of hours, and is doing a lot of extra work. So my questions are: Can you make dh_install do a hardlink copy (like cp -l) to save time? (AFAICT from the man page, no.) Maybe I should just get it to do this in the Makefile? (That's gonna b e big Makefile.) Can you make debuild only rebuild .debs that need rebuilding? Or specify which .debs to rebuild? Is my approach completely stupid? Should I break each of the toolboxes into its own source package too? (I'll have to do some silly templating or something, because there's hundreds of them. :/)

    Read the article

  • Should developers be involved in testing phases?

    - by LudoMC
    Hi, we are using a classical V-shaped development process. We then have requirements, architecture, design, implementation, integration tests, system tests and acceptance. Testers are preparing test cases during the first phases of the project. The issue is that, due to resources issues (*), test phases are too long and are often shortened due to time constraints (you know project managers... ;)). So my question is simple: should developers be involved in the tests phases and isn't it too 'dangerous'. I'm afraid it will give the project managers a false feeling of better quality as the work has been done but would the added man.days be of any value? I'm not really confident of developers doing tests (no offense here but we all know it's quite hard to break in a few clicks what you have made in severals days). Thanks for sharing your thoughts. (*) For obscure reasons, increasing the number of testers is not an option as of today. (Just upfront, it's not a duplicate of Should programmers help testers in designing tests? which talks about test preparation and not test execution, where we avoid the implication of developers)

    Read the article

  • External Full HD monitor and Virtual Desktop Size

    - by Stefan
    I have two FullHD monitors attached to my ATI graphics card [2]. The resolution of both of them is detected properly without any modifications to /etc/X11/xorg.conf. I can run both of them in clone mode. However, when I try to run them next to each other, I got the following error: The selected configuration for displays could not be applied. If tried to fix this according to [1]. My xorg.conf now looks like this: Section "Module" Load "glx" EndSection Section "Screen" Identifier "Default Screen" DefaultDepth 24 SubSection "Display" # The 1088 is the smallest multiple of 32 >= 1088 # see manpages Virtual 1920 1088 EndSubSection EndSection This does not seem to be parsed properly. After restarting X, I cannot set resolutions beyond 1600 or so any more. /var/log/Xorg.0.log gives: [ 15.676] (II) fglrx(0): Not using mode "1920x1080" (width too large for virtual size) [ 15.676] (II) fglrx(0): Not using mode "1680x1050" (width too large for virtual size) Are my modifications syntactically incorrect? According to the man page, it should be fine. Any ideas? OS: Ubuntu 11.10 64bit [1] http://askubuntu.com/a/75546/5023 [2] 01:00.0 VGA compatible controller: ATI Technologies Inc Juniper [Radeon HD 5700 Series]

    Read the article

  • How do I make apt-get instal commands not display every package that is installing?

    - by rajlego
    When I install something on terminal, it often shows me a few things for status. For one, it shows download rate (which is fine). However, when I install something, it can display Unpacking libgranite2:amd64 (0.3.0~r732+pkg64~ubuntu0.3.1) ... Selecting previously unselected package slingshot-launcher. Preparing to unpack .../slingshot-launcher_0.7.6.1+r421+pkg32~ubuntu0.3.1_amd64.deb ... Unpacking slingshot-launcher (0.7.6.1+r421+pkg32~ubuntu0.3.1) ... Selecting previously unselected package contractor. Preparing to unpack .../contractor_0.3.1~r136+pkg22~ubuntu0.3.1_amd64.deb ... Unpacking contractor (0.3.1~r136+pkg22~ubuntu0.3.1) ... Selecting previously unselected package apport-hooks-elementary. Preparing to unpack .../apport-hooks-elementary_0.1-0~35~saucy1_all.deb ... Unpacking apport-hooks-elementary (0.1-0~35~saucy1) ... Processing triggers for hicolor-icon-theme (0.13-1) ... Processing triggers for libglib2.0-0:amd64 (2.40.0-2) ... Processing triggers for man-db (2.6.7.1-1) ... Setting up libgranite-common (0.3.0~r732+pkg64~ubuntu0.3.1) ... Setting up libgranite2:amd64 (0.3.0~r732+pkg64~ubuntu0.3.1) ... Setting up slingshot-launcher (0.7.6.1+r421+pkg32~ubuntu0.3.1) ... Setting up contractor (0.3.1~r136+pkg22~ubuntu0.3.1) ... Setting up apport-hooks-elementary (0.1-0~35~saucy1) ... Processing triggers for libc-bin (2.19-0ubuntu6) .. I would rather that not show up. I only want to see download rate, not all that other stuff. How do I do this? EDIT: I would also like the jargon to be stored somwehre else if something goes wrong, or for the jargon to just be expanable on terminal.

    Read the article

  • Never Bet Against the Impossible

    - by BuckWoody
    My uncle used to say “If a man tells you that his car squirts milk in his eye when you lift the hood, don’t bet against that. You’ll end up with milk in your eye.” My friend Allen White tells me this is taken from a play (and was said about playing cards), but I think the sentiment holds, even in database work. I mentioned the other day that you should allow the other person to talk and actively listen before you propose a solution. Well, I saw a consultant “bet against the impossible”  the other day – and it bit her. She explained to the person telling her the problem that the situation simply couldn’t exist that way, and he proceeded to show her that it did. She got silent, typed a few things, muttered a little, and then said “well, must be something else.” She just couldn’t admit she was wrong. So don’t go there. If someone explains a problem to you with their database, listen with purpose, and then explore the troubleshooting steps you know to find the problem. But keep your absolutes to yourself. In fact, I have a friend that has recently sent me one of those. He connects to a system with SQL Server Management Studio (SSMS) version 2008 (if I recall correctly) and it shows a certain version number of the target system in the connection tab. Then he connects to it using SSMS 2008 R2 and gets a different number. Now, as far as I know, we didn’t change the connection string information, and that’s provided by the target system, so this is impossible. But I won’t tell him that. Not until I look a little more. :) Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • Can I set up samba so it automatically allows all the local usernames and passwords?

    - by dialer
    I have set up samba like this (this is the complete smb.conf): [global] log file = /var/log/samba/log log level = 2 security = user [homes] browsable = false read only = no valid users = %S I'd like to enable every user on server to access their home directories, but for some unknown reason only my 'administrator' account can do so. (I have done that with ftp before, but now smb is also needed). When I try to smbclient -L localhost -U [user], I get NT_STATUS_LOGON_FAILURE, except with the administrator (which is the user created during the ubuntu installation, not root). The samba log file says NT_STATUS_NO_SUCH_USER: [2012/04/04 20:26:02.081454, 2] smbd/reply.c:554(reply_special) netbios connect: name1=LOCALHOST 0x20 name2=DIALER-X 0x0 [2012/04/04 20:26:02.081733, 2] smbd/reply.c:565(reply_special) netbios connect: local=localhost remote=dialer-x, name type = 0 [2012/04/04 20:26:02.087200, 2] auth/auth.c:314(check_ntlm_password) check_ntlm_password: Authentication for user [public] - [public] FAILED with error NT_STATUS_NO_SUCH_USER I suspect that I have to manually create samba users, but the man pages state that If the client has passed a username/password pair and that username/password pair is validated by the UNIX system's password programs, the connection is made as that username. To me that sounds like as long as the provided username/password is a valid login on the server, it should work. Am I missing something totally obvious? I don't want / can't afford to manually update the samba users and passwords to match the server's. 11.10

    Read the article

  • Oracle Virtual Desktop Infrastructure

    - by Fat Bloke
    A lot of the recent blog entries here have been about Oracle VM VirtualBox, possibly the coolest personal desktop virtualization product known to man. Deploying VirtualBox on your PC or Mac lets you run many virtual desktops at the same time to one user, you. But did you know that VirtualBox can also power an Enterprise-scale virtual desktop deployment too, delivering many desktops to many users?  As part of another Oracle product, Oracle Virtual Desktop Infrastructure (VDI), VirtualBox can run your Windows, Linux or Solaris desktops on servers located in the datacenter. Oracle VDI orchestrates the whole deal by looking after : creating or cloning the virtual desktops from a master template; managing the lifecycle of the desktops (create, start, suspend, resume, stop, delete); assigning which users get which desktops;  delivering easy and fast access to these virtual desktops from almost any device, such as existing PCs or Macs, iPads, or specially designed Sun Ray client devices too; load balancing and session management of all of this.  Architecturally the solution looks something like this: This is an increasingly hot area of the IT landscape, so the Fat Bloke has decided to create a new blog category (VDI) and dedicate a few blog entries to look into this in a bit more detail over the next few weeks. Watch this space... - FB 

    Read the article

  • NFS mount of /var/www to OS X

    - by ploughguy
    I have spent 2 hours trying to create an NFS mount from my Ubuntu 10.04 LTS server to my OS X desktop system. Objective: three way file compare between the code base on the Mac, the development system on the local Linux test system, and the hosted website. The hosted service uses cpanel so I can mount a webdisk - easy as pie - took 10 seconds. The local Ubuntu box, on the other hand - nothing but pain and frustration. Here is what I have tried: In File Browser, navigate to /var/www/site and right-click. Select share this folder. Enter sharename wwwsite and a comment. Click button "Create Share". Message says - you can only share file systems you own. There is a message on how to fix this, but the killer is that this is sharing by SMB. It will change the LFs to CR-LFs which will affect the file comparison. So forget this option. In a terminal window, run shares-admin (I have not been able to convince it to give me the "Shared Folders" option in the System Administration window - Maybe it is somewhere else in the menu, but I cannot find it) define an NFS export. Enter the path /var/www/site, select NFS enter the ip address of the iMac and save. On the mac, try to mount the file system using the usual methods - finder, command line "mount" command - not found. Nothing. Tried restarting the linux box in case there is a daemon that needs restarting - nothing. So I have run out of stuff to do. I have tried searching the documentation - it is pretty basic. The man page documentation is as opaque as ever. Please, oh please, will someone help me to get this @38&@^# thing to work! Thanks for reading this far... PG.

    Read the article

  • How do I add a boot from cd option to yaboot?

    - by Sergiu
    So I'm dual-booting Ubuntu 12.04.1 on my iMac G5 powepc alongside Mac OS X and I want to add a boot cd option to yaboot because I'm trying to boot a scratched Mac OS X installation DVD that takes a while to read and the frst bootstrap moves on too fast. How do I edit the timeout for the first bootstrap anyways? So, my main question is, how do I add a cd booting option to yaboot and then, how doI boot it? The devalias from OpenFrmware tells me that 1 have 2 cd-rom instaled, on is /ht/pci@3/ata-6/disk@0 and the other on ends with a 1 instead of a zero. These are the contents of my yaboot.conf file: yaboot.conf generated by the Ubuntu installer run: "man yaboot.conf" for details. Do not make changes until you have!! see also: /usr/share/doc/yaboot/examples for example configurations. For a dual-boot menu, add one or more of: bsd=/dev/hdaX, macos=/dev/hdaY, macosx=/dev/hdaZ boot="/dev/disk/by-id/scsi-SATA_ST3160023AS_5MT1GCWA-part2" device=/ht@0,f2000000/pci@3/k2-sata-root@c/@0/@0 partition=4 root="UUID=798a048f-ee48-49e0-bba3-111aed8dee04" timeout=12000 install=/usr/lib/yaboot/yaboot magicboot=/usr/lib/yaboot/ofboot enablecdboot macosx="/dev/disk/by-id/scsi-SATA_ST3160023AS_5MT1GCWA-part3" image=/boot/vmlinux label=Linux read-only initrd=/boot/initrd.img append="quiet splash" What do I add here so that yaboot will boot from my cd in like 3 minutes after startup? Thanks!

    Read the article

  • How can one find software development work that involves directly the final end user?

    - by RJa
    I've worked in software development for 15 years and, while there have been signficant personal achievements and a lot of experience, I've always felt detached from the man/woman-on-the-street, the every day person, how it affects their lives, in a number of ways: the technologies: embedded software, hidden away, stuff not seen by the everyday person. Or process technology supporting manufactured products the size of the systems, meaning many jobs, divided up, work is abstract, not one person can see the whole picture the organisations: large, with departments dealing with different areas, the software, the hardware, the marketing, the sales, the customer support the locations and hours: out-of-town business parks away from the rest of society, fixed locations, inflexible: 9-5 everyday This to me seems typical of the companies I worked for and see elsewhere. Granted, there are positives such as the technology itself and usually being among high calibre co-workers, but the above points frustrate me about the industry because they detach the work from its meaning. How can one: change these things in an existing job, or compensate for them? find other work that avoids these and connects with the final end user? Job designs tend to focus on the job content and technical requirements rather than how the job aims to fulfil end user needs, is meaningful.

    Read the article

  • how to solve this problem

    - by Surbir
    root@me-desktop:~# sudo apt-get install aircrack-ng Reading package lists... Done Building dependency tree Reading state information... Done The following NEW packages will be installed: aircrack-ng 0 upgraded, 1 newly installed, 0 to remove and 446 not upgraded. 1 not fully installed or removed. Need to get 1,579kB of archives. After this operation, 2,843kB of additional disk space will be used. Get:1 http://archive.ubuntu.com/ubuntu/ maverick/universe aircrack-ng i386 1:1.1-1 [1,579kB] Fetched 1,579kB in 1min 9s (22.7kB/s) Selecting previously deselected package aircrack-ng. (Reading database ... 520739 files and directories currently installed.) Unpacking aircrack-ng (from .../aircrack-ng_1%3a1.1-1_i386.deb) ... Processing triggers for man-db ... Setting up linux-image-3.0.1-030001-generic (3.0.1-030001.201108060905) ... Running depmod. update-initramfs: Generating /boot/initrd.img-3.0.1-030001-generic Warning: No support for locale: en_US.utf8 Examining /etc/kernel/postinst.d. run-parts: executing /etc/kernel/postinst.d/dkms 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic run-parts: executing /etc/kernel/postinst.d/initramfs-tools 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic run-parts: executing /etc/kernel/postinst.d/nvidia-common 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic run-parts: executing /etc/kernel/postinst.d/pm-utils 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic run-parts: executing /etc/kernel/postinst.d/update-notifier 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic run-parts: executing /etc/kernel/postinst.d/zz-update-grub 3.0.1-030001-generic /boot/vmlinuz-3.0.1-030001-generic exec: 15: update-grub: not found run-parts: /etc/kernel/postinst.d/zz-update-grub exited with return code 2 Failed to process /etc/kernel/postinst.d at /var/lib/dpkg/info/linux-image-3.0.1-030001-generic.postinst line 1010. dpkg: error processing linux-image-3.0.1-030001-generic (--configure): subprocess installed post-installation script returned error exit status 2 Setting up aircrack-ng (1:1.1-1) ... Errors were encountered while processing: linux-image-3.0.1-030001-generic E: Sub-process /usr/bin/dpkg returned an error code (1) root@me-desktop:~#

    Read the article

  • How do I make apt-get install commands not display every package that is installing?

    - by rajlego
    When I install something on terminal, it often shows me a few things for status. For one, it shows download rate (which is fine). However, when I install something, it can display Unpacking libgranite2:amd64 (0.3.0~r732+pkg64~ubuntu0.3.1) ... Selecting previously unselected package slingshot-launcher. Preparing to unpack .../slingshot-launcher_0.7.6.1+r421+pkg32~ubuntu0.3.1_amd64.deb ... Unpacking slingshot-launcher (0.7.6.1+r421+pkg32~ubuntu0.3.1) ... Selecting previously unselected package contractor. Preparing to unpack .../contractor_0.3.1~r136+pkg22~ubuntu0.3.1_amd64.deb ... Unpacking contractor (0.3.1~r136+pkg22~ubuntu0.3.1) ... Selecting previously unselected package apport-hooks-elementary. Preparing to unpack .../apport-hooks-elementary_0.1-0~35~saucy1_all.deb ... Unpacking apport-hooks-elementary (0.1-0~35~saucy1) ... Processing triggers for hicolor-icon-theme (0.13-1) ... Processing triggers for libglib2.0-0:amd64 (2.40.0-2) ... Processing triggers for man-db (2.6.7.1-1) ... Setting up libgranite-common (0.3.0~r732+pkg64~ubuntu0.3.1) ... Setting up libgranite2:amd64 (0.3.0~r732+pkg64~ubuntu0.3.1) ... Setting up slingshot-launcher (0.7.6.1+r421+pkg32~ubuntu0.3.1) ... Setting up contractor (0.3.1~r136+pkg22~ubuntu0.3.1) ... Setting up apport-hooks-elementary (0.1-0~35~saucy1) ... Processing triggers for libc-bin (2.19-0ubuntu6) .. I would rather that not show up. I only want to see download rate, not all that other stuff. How do I do this? EDIT: I would also like the jargon to be stored somwehre else if something goes wrong, or for the jargon to just be expanable on terminal.

    Read the article

  • Agressive Auto-Updating?

    - by MattiasK
    What do you guys think is best practice regarding auto-updating? Google Chrome for instance seems to auto-update itself as soon as it get's a chance without asking and I'm fine with it. I think most "normal" users benefits from updates being a transparent process. Then again, some more technical users might be miffed if you update their app without permission, as I see it there's 3 options: 1) Have a checkbox when installing that says "allow automatic updates" 2) Just have a preference somewhere that allows you to "disable automatic updates" so that you have to "check for updates manually" I'm leaning towards 2) because 1) feels like it might alienate non-technical users and I'd rather avoid installation queries if possible. Also I'm thinking about making it easy to downgrade if an upgrade (heaven forbid) causes trouble, what are your thoughts? Another question, even if auto-updates are automatically, perhaps they should be announced. If there's new features for example otherwise you might not realize and use them One thing that kinda scares me though is the security implications, someone could theorically hack my server and push out spyware/zombieware to all my customers. It seems that using digital signatures to prevent man-in-the-middle attacks is the least you could do otherwise you might be hooked up to a network that spoofs the address of of update server.

    Read the article

  • Securing a Cloud-Based Data Center

    - by Orgad Kimchi
    No doubt, with all the media reports about stolen databases and private information, a major concern when committing to a public or private cloud must be preventing unauthorized access of data and applications. In this article, we discuss the security features of Oracle Solaris 11 that provide a bullet-proof cloud environment. As an example, we show how the Oracle Solaris Remote Lab implementation utilizes these features to provide a high level of security for its users. Note: This is the second article in a series on cloud building with Oracle Solaris 11. See Part 1 here.  When we build a cloud, the following aspects related to the security of the data and applications in the cloud become a concern: • Sensitive data must be protected from unauthorized access while residing on storage devices, during transmission between servers and clients, and when it is used by applications. • When a project is completed, all copies of sensitive data must be securely deleted and the original data must be kept permanently secure. • Communications between users and the cloud must be protected to prevent exposure of sensitive information from “man in a middle attacks.” • Limiting the operating system’s exposure protects against malicious attacks and penetration by unauthorized users or automated “bots” and “rootkits” designed to gain privileged access. • Strong authentication and authorization procedures further protect the operating system from tampering. • Denial of Service attacks, whether they are started intentionally by hackers or accidentally by other cloud users, must be quickly detected and deflected, and the service must be restored. In addition to the security features in the operating system, deep auditing provides a trail of actions that can identify violations,issues, and attempts to penetrate the security of the operating system. Combined, these threats and risks reinforce the need for enterprise-grade security solutions that are specifically designed to protect cloud environments. With Oracle Solaris 11, the security of any cloud is ensured. This article explains how.

    Read the article

  • Aggressive Auto-Updating?

    - by MattiasK
    What do you guys think is best practice regarding auto-updating? Google Chrome for instance seems to auto-update itself as soon as it get's a chance without asking and I'm fine with it. I think most "normal" users benefits from updates being a transparent process. Then again, some more technical users might be miffed if you update their app without permission, as I see it there's 3 options: 1) Have a checkbox when installing that says "allow automatic updates" 2) Just have a preference somewhere that allows you to "disable automatic updates" so that you have to "check for updates manually" I'm leaning towards 2) because 1) feels like it might alienate non-technical users and I'd rather avoid installation queries if possible. Also I'm thinking about making it easy to downgrade if an upgrade (heaven forbid) causes trouble, what are your thoughts? Another question, even if auto-updates are automatically, perhaps they should be announced. If there's new features for example otherwise you might not realize and use them One thing that kinda scares me though is the security implications, someone could theorically hack my server and push out spyware/zombieware to all my customers. It seems that using digital signatures to prevent man-in-the-middle attacks is the least you could do otherwise you might be hooked up to a network that spoofs the address of of update server.

    Read the article

  • There's Not an App for That (Yet)

    - by Mark Hesse
    With an earlier-than-normal departure this morning to avoid the stalemate known as traffic congestion, I suddenly realized what I had failed to grab on my way out the door...  my company ID badge.  Unfortunately, at the time of my epiphany, I was far enough into commuter no-man's land where turning back would completely negate my early departure and increase my overall drive time exponentially.  Not being one to retrace my steps, I decided to press on. Upon arrival at the office and with an hour to go before a security guard would be on duty, I started thinking about the number of times I had forgotten my ID vs. the number of times I had forgotten my phone.  While rare on both accounts, my ID was most likely the missing artifact. I then wondered why there isn't an app for my smartphone that allows me to verify my credentials with my employer and then, provided with a secure token for the day, have the ability to access my building's card entry system.  On many levels, this seems much more secure than an ID card which can be lost, stolen or even forged and then used simply by tailgating into and around buildings at facilities where card scanning can generally be avoided.   As it turns out, another building on the campus has 24 x 7 guard coverage, so I was able to gain access in a relatively short time and secure a temporary ID badge.  Once inside and online, a quick internet search on the subject of smartphone badge access shows that efforts are underway to do exactly what I was thinking needed to be done. Having not spent any time studying about the technology, I discovered that it relies on Near Field Communications (NFC) enabled smartphones (of which, mine does not provide).  The only other option would require modifications to the security infrastructure to support alternative authentication technologies, such as barcode readers, which would be extremely costly to implement. For now, my best option is to put my corporate ID under my car keys... 

    Read the article

  • Java vs. C# - Productivity perspective

    - by Edin Dazdarevic
    If you have a number of years experience in working with JAVA and a number of years experience in working with C# and .NET, I would value your opinion on software development productivity differences between these two environments. One of our customers is considering to technically replace their existing software solution. As the replacement will require approx. 10 - 15 man years work, a choice for JAVA or .NET, based on productivity differences between the them, may significantly influence the investment required and time-to-market. Would you be able to provide us, based on your honest and expert opinion, an indication of software development productivity differences between JAVA and C#/.NET? I would prefer to receive an answer as follows: My experience is based on X years experience working with JAVA and X years experience working with C#/.NET. JAVA is X% more productive then C#.NET or C#/.NET is X% more productive than JAVA if you take the the following into account . . . . . . . Thanks

    Read the article

  • Enjoy Cloud Odyssey The Oracle Movie

    - by Naresh Persaud
    If you attended Open World you may have seen the promotions for a new movie produced by Oracle. The movie is called Cloud Odyssey and it chronicles the journey of a hero to the cloud. The movie is an animated sci-fi adventure. This movie will be played at Oracle events around the world so you may soon get an invite to attend. Interesting approach to telling the cloud story. For many IT organizations, the journey to the cloud is a major initiative for end users. I am sure Homer would be proud. In fact perhaps if it is successful, I am hopeful we may see a cloud Iliad.  Below, I have embedded a trailer to the movie for your viewing pleasure. While it clearly is not the next Iron Man, it is intriguing. Hope you enjoy.  &amp;amp;amp;amp;amp;lt;span id=&amp;amp;amp;amp;amp;quot;XinhaEditingPostion&amp;amp;amp;amp;amp;quot;&amp;amp;amp;amp;amp;gt;&amp;amp;amp;amp;amp;lt;/span&amp;amp;amp;amp;amp;gt;

    Read the article

  • Autojump in 12.04 doesn't work

    - by hnasarat
    https://launchpad.net/ubuntu/+source/autojump I installed with apt-get, checked out the man page and added . /usr/share/autojump/autojump.sh to my .bashrc, like it says. When I cd around the filesystem, nothing gets added to ~/.local/share/autojump. I then tried adding . /usr/share/autojump/autojump.bash, but that didn't work either. autojump -a ~/Dropbox properly creates a file ~/.local/share/autojump/autojump.txt, but running j Drop < TAB > doesn't autocomplete to j ~/Dropbox/ as it should. However, j < TAB > does autocomplete to j ~/Dropbox. I know my bash-completion is working since it works for git, dd, and others. I know there's a newer version in the repositories set for Quantal. Perhaps that would work? I don't know how to install that version though. I've used autojump with mac homebrew (and it installed without any issue), so I know there is missing functionality. In general I'm really annoyed that I can't get this working...I've spent hours on it! Needless to say, help would be very appreciated.

    Read the article

  • New mainboard with 890GX chipset disables lightdm, even when using old graphics card instead of onboard graphics, startx works (Xubuntu 12.04)

    - by user99250
    I am trying to migrate an installed Xubuntu 12.04 to a new mainbaord with 890GX chipset. The Chipset has a built-in Radeon HD 4290 graphics. The system boots, but X won't start. The most suspicious message in Xorg.0.log is "ddxSigGiveUp: Closing log". When searching for this message, I found some answers like "remove your xorg.conf" (but there is none on my system). Or bug reports for fglrx (but that's not installed on my system). Or NVidia-related questions ... Interestingly, "startx" succeeds in opening a basic XFCE session. Then, I tried to disable the onboard graphics in the BIOS setup and use the old PCIe graphics card (Radeon HD 5450). No change. I don't think I can just blacklist a module, because the graphics card and the onboard graphics are covered by the same module. At the moment, use the free radeon drivers, not the restricted fglrx driver. If possible, I would like to stay on the free driver for two reasons: The fglrx driver from the ubuntu packages fails to build the kernel module. In past, I had bad experiences with the fglrx driver and changing screen config with RandR. When I connect the harddisk and the graphics card to my old mainboard, everything works again. This means, I have not screwed up my system configuration wile installing and removing the fglrx drivers. When I ordered this mainboard, I thought the 890GX is old enough to be supported and if not, I could still use my graphics card as backup solution. But without even the backup soluton, I'm screwed ... Any ideas ? Thanks and Regards, Kubuntu-Man (now switched to Xubuntu)

    Read the article

  • VCS for single user using file sync service

    - by StackUnder
    I'm trying to setup a version control for my one man project. My project files are in sync thanks to live mesh (but I could be using dropbox for that matter), between my laptop, my home pc and my office pc. I'm now using Netbeans with local file history. Sometimes it helps to revert to a previous state of one file. But imagine a situation when multiple files have problems. Correct me if I'm wrong but I would have to go to every file and revert to previous "safe" state. I don't like this approach, so I'm considering using a version control between SVN and GIT. I have some previous experience with SVN (TortoiseSVN) and I know that I can create a file:// repo. So, what a want to do is setup a VCS inside my synced folder just to have the ability to "revert" to a previous version if something goes wrong. Since everything's been synced to all computers, I wouldn't ever need to run an update. The file tree organization would be the following: C:...\SyncedFolder\MyProject\ Inside MyProject folder are all the project files plus a directory that has SVN or GIT info of my project (the repo/master). What VCS is best for this situation: SVN or GIT? Does SVN need to store all files from HEAD revision, thus "duplicating" all my project inside my synced folder? Does GIT eliminates this problem? Is this the best approach?

    Read the article

  • Pointing non-www to a spcific sub-directory

    - by Ben Sinclair
    I might be going about this all wrong so let me know if I am. I am creating software that allows people to sign up and have their own sub-domain on my website. So say my website is ben.com, they could have their own sub-domain called juice.ben.com. When they type their sub-domain juice.ben.com in their address bar, it will load the contents in a root directory. I have also set-up a .htaccess redirect to redirect www.ben.com to ben.com. Not sure if this matters with my question but I thought I'd mention it. Ok, so basically what I think I need to do is put the software they they've signed up to in the root directory. So when someone goes to juice.ben.com, they will be pointed to the root directory (I beleive I cans et-up wild card sub-domains with my host) and the software will then analyse their sub-domain and then display their account. Now, if someone just types in ben.com into their browser, I want it to show the contents of the ben.com/_website/ folder but still show in the address bar that they are still in the root directory. Hopefully I am making sense :) Is this possible with htaccess? If so, what do I need to do?

    Read the article

  • Using IsolatedStorage on a IIS server

    - by JoeBilly
    I'am a bit confusing about the use of Isolated Storage on an IIS server. I understand the goal of Isolated Storage : provides a safe place to store data with no worry about how and where is this place. Since Isolated Storage has a by-user and by-assembly approach, I'am not to wild about using it on a IIS server where applications have almost their own identity. I haven't really seen the interest of impersonating a web application and almost never seen impersonated web applications myself but this is my point of view. Using Isolated Storage on a server mean : Using Isolated stores in \Documents and Settings\<user>\ Which mean \Documents and Settings\Default User\ when the application pool is owned by Local System or Network Services I guess Which also mean Write rights on this folder for Local System or Network Services Using of impersonation Regarding a web application (logic), these ideas are confusing me... Document and Settings ? Default User ? Enable impersonation just for storage ? No control about storage on server ? Uh ? And then I'am a front of a dilema : use System.IO.Packaging (with Isolated Storage inside) on web applications or find an alternative ? Am I wrong in my approach ? Did I miss something ? Any point of view is appreciated and an explanation about the Isolated Storage with IIS philosophy could be an anwser. Thanks !

    Read the article

  • Is there a Standard or Best Practice for Perl Programs, as opposed to Perl Modules?

    - by swestrup
    I've written any number of perl modules in the past, and more than a few stand-alone perl programs, but I've never released a multi-file perl program into the wild before. I have a perl program that is almost at the beta stage and is going to be released open source. It requires a number of data files, as well as some external perl modules -- some I've written myself, and some from CPAN -- that I'll have to bundle with it so as to ensure that someone can just download my program and install it without worrying about hunting for obscure modules. So, it sounds to me like I need to write an installer to copy all the files to standard locations so that a user can easily install everything. The trouble is, I have no idea what the standard practice would be for this. I have found lots of tutorials on perl module standards, but none on perl program standards. Does anyone have any pointers to standard paths, installation proceedures, etc, for perl programs? This is going to be complicated by the fact that the program is multi-platform. I've been testing it in Linux, but its designed to work equally well in Windows.

    Read the article

< Previous Page | 55 56 57 58 59 60 61 62 63 64 65 66  | Next Page >