Windy Weather » Linux

Ubuntu Karmic Koala Windows File Sharing

Darrell — Fri, 22 Jan 2010 20:42:41 +0000

So, when you install Karmic Koala [9.10], you still don’t get Windows File sharing [SAMBA], and setting it up is still a few undocumented and obscure steps. I covered this before for Gutsy, but things are a little different with Karmic, so I’ve redone the screen shots and directions.

If you could Google for “Ubuntu Samba Install” and get a page that told you what you wanted to know, then I wouldn’t do this post, but you don’t get a good page. This page is not that helpful. It looks it was written in the dark ages and it does not configure WINS so that you can PING the other systems on your network.

So here’s the real scoop, which is a modified version of this post I did for Gutsy.

First you need to install WINBIND. So use the package manager at SYSTEM > Administration > Synaptic Package Manager. Search for WINBIND and mark for install. This picture shows what you need to have – or at least what I have installed. I think if you search successively for WINBIND, SAMBA, SMBCLIENT and SMB you will find these. Many of the things in this picture are probably dependencies that will be pulled in automatically.

Now you need to edit the /etc/nsswitch.conf file to change the order to allow wins to work.

This change allows winbind to find the other computers on the network. This allows PING to work, which allows you to diagnose issues, and allows SAMBA to find other systems on the network using the wins protocol. You don’t need to restart anything or reboot after you edit this file. When this is working, you should get the following when you use System > Administration > Network Tools > Ping:

Ok, the next step is to configure SAMBA by editing the /etc/samba/smb.conf file. The system that I upgraded worked fine after the upgrade from Gutsy to Karmic because this file was preserved, and because the upgrades [ 4 of them] all preserved the packages already installed. But if you are installing a new system, then you will need to change the smb.conf file since it does not work out of the box, as far as I can tell. Here’s a terminal session that shows what to do to edit the file. I’m not going to describe the whole Sudo thing, you can find that elsewhere.

Notice that Samba needs a restart after any changes to smb.conf. I’m not going to go through all the details like I did last time, you can refer to the previous post. Here’s a copy of the file that I’m using now.

smb_conf

There is one more thing to do. Since we are using security = user with samba, we must set a samba password for every account that we are going to use with samba. I believe this results from the fact that windows users md4 for password hashing, and passes this hash across the network to authenticate users. Since linux [unix] uses a stronger password hash, you must save the MD4 hash of the password in a samba database to allow authentication for samba. Here’s the process from the other post:

darrell@squall-ubuntu:/etc/samba$ sudo smbpasswd -a darrell [sudo] password for darrell: New SMB password: Retype new SMB password: darrell@squall-ubuntu:/etc/samba$

Ignore the following material, it is fixed by the above smbpasswd fix.

There are still problems that I cannot solve:

With this configuration I get two way file sharing with Vista x86 x64 and Win xp on the system that was upgraded. But when I installed a new system, and followed these steps, including using exactly this smb.conf file, the newly installed system would not allow Vista x86 or x64 clients to access Ubuntu. Win XP as the client worked just fine. Ubuntu as the client worked for Vista x86, x64 and Win XP. Read only. I only allow read-only file shares on my network to avoid the spread of worms should I get them.

Looking into a Vista x64 Machine

Here’s an example of an error when trying to connect to Ubuntu from Vista.

This is the network as seen from a Vista x86 machine. You can see that both Ubuntu machines, Squall-Ubuntu and Blue-Diamond are visible.

Network as Seen from Vista x86

You can connect to Squall-Ubuntu, but not to Blue-Diamond.

It doesn’t matter whether you put in darrell, or blue-diamond\darrell, you get the same error. But if you connect to squall-ubuntu, which is the upgraded machine, then it works just fine. I have no clue what the differences are between the configurations. Recall that I’m using the same smb.conf file on both machines and that both have the same nsswitch.conf file.

No clue what’s wrong.

But the Ubuntu machines can see all the Windows machines so I can pull files from them if I like. I pulled the pictures from those machines and did this post from Squall-Ubuntu.

So Karmic can play nice with windows, almost. But it’s still not that way out of the box.

- windy

Karmic Koala – File Sharing, and Display Size Glitch

Darrell — Thu, 21 Jan 2010 20:01:19 +0000

I just upgraded my Ubuntu Linux system to Karmic Koala [ 9.10 ] from Gutsy Gibbon [7.04]. As you can tell, it had been some time since I had upgrade. The good news is that the upgrades, through 8.04, 8.10, 8.04, to 9.10, went without a hitch, although they took over 3 hours.

And network file sharing, SAMBA or SMB, works without a hitch with no modifications to the built in configuration. Hurray!!!. You may remember a previous post about Ubuntu Not Playing Nice that showed the modifications that I had to do to BIND and to SMB configuration to get things to work. UPDATE: file sharing worked because I had fixed it in Gutsy. It still DOES NOT WORK OUT OF THE BOX with Karmic Koala.

See this post for how to start to make it work with Karmic [9.10].

Well everything works just fine now. No changes to any of the default configuration files. As you visit the first sites, I had to allow Samba to access the KeyRing and enter passwords, but after that all is well. I use the same login for all my machines, so I only had to enter that login once for it to work everywhere. I have Win XP, Vista x86 and Vista x64 machines so this is good news for almost everyone. BTW, it works to another machine running Gutsy too, which I’m in the process of upgrading.

But there was a small glitch when I did the last upgrade, but I’ll get to that in a moment.

The upgrade process takes you though each of the upgrades in sequence. No skipping ahead. So it takes a while. There are dialogs that tell you about the changes you may have made to configuration files for some components, like SAMBA, so you have to watch for those since they stop the upgrade. At first I keep my changes, but finally decided to see if they had fixed problems and said REPLACE to take the new files. There were only three of those as I recall, SAMBA and a couple of others, like Internet Time and one other that was so obscure that I don’t recall having changed it.

When the upgrades finally finished, the display was left in a strange state. The desktop was too large for the screen, and dragging the mouse panned the desktop around on the screen. This computer has an NVIDIA display controller, and apparently there is a little glitch or two to be worked out with that integration. Here’s a picture of the entire Desktop image as captured with Screenshot.

If you try to set the display size you get an error.

If you say YES at this point, which is what is indicated, then you go to the NVIDIA panel and you can’t fix the panning problem. Here’s the NVIDIA panel.

But this is set correctly to 1600 x 1050 [ which is my monitor size.] But… that doesn’t fix the panning issue.

There is another error with this panel that I don’t understand.

If you save the configuration, you get an error.

Not sure what’s up with that.

So the right way to fix this is to say YES to the first dialog about the graphics driver support and you get the built-in display dialog:

This was set incorrectly to 1900×1200, which caused the panning. By setting the resolution to 1600×1050, all is well. Minor glitch.

By the way, I enabled visual effects in System > Preferences > Appearance > Visual Effects > Extra, and the waving windows are way cool.

The reason for upgrading my Ubuntu is so that I can do Safe Online Banking with no fear of Trojans stealing my banking information. I’m going to be doing all my online banking using Ubuntu now.

Have a good one.

- Windy

Media Servers for a PS3

Darrell — Sun, 23 Mar 2008 05:21:17 +0000

The PS3 is probably the best, and reasonably priced, Blu-Ray player on the market. Now that the HD format wars are over, it was time to get a Blu-Ray player

All the Windows hype about the “Media Editions” of Windows was never very clear. Apparently the Media Edition contains the features necessary to serve various files to the other entertainment systems in the home. Music and video players can play files from your desktop harddrive via these servers. Now with a new PS3, Media Services is becoming more useful. It will be interesting to play photos and videos from a server. The PS3 only has a 40GB harddrive and the file management features of the PS3 are very primitive.

Summary

TwonkyMedia Server is the best one that I’ve found. Very reasonable price – $39.95US. It runs on both Windows [Vista x64] and Ubuntu Linux with no problem. It is very efficient using very little memory and processor time. It works just fine on my little Blue-Diamond 800MHz VIA system.

Test Cases

The test cases are modest. About 860 images – jpgs 1024×768 on the PC and 1280×1024 on the Linux box. About 250 mp3 files. The images all display on the PS3 from a USB flash key. Almost all the music files play from a data DVD when inserted into the PS3. Some show “Corrupted File” on the PS3 for an unknown reason. Most of the mp3 files play with no problems using WinAMP on a Windows PC. As of today, the PS3 is running version 2.17 of the firmware.

Update: I’ve discovered the source of some, if not all, of the Corrupted File indications.

Using MP3Tag V2.40 I’ve discovered that a few of the files had very strange meta data, and were in an unsupported format. Click on the pic above to see the full size view. Note that the highlighted files have strange bitrates and frequencies. I’m not sure whether this causes the server to trip up, or just the PS3.
After retesting Twonky and gMediaServer after cleaning out the strange files allowed Twonky to sometimes serve the files, but often it tripped up with a 2103 protocol error and basically stopped working after that. gMediaServer still would not serve any of the sanitized mp3 files.

Nero Media Home – Ultra 8

A google search for “PS3 media servers” indicates that Nero Media Home is a media server. But I suggest you don’t waste your time with it. There are several reasons:

It is very resource intensive, shall we say, taking over 60MB of your precious ram to run the server. This burden requires an indexing service, which is a constant load on the system regardless of whether you are using the server or not.
It’s not fast in it’s response to serving to the PS3. It takes a very long time to get the list of items to the PS3.
It causes protocol error reports on the PS3. “DLNA Protocol Error 80710092″ whatever that is.
The NMH crashes and hangs quite often, esp when you stop and start the server.
It won’t serve mp3s. You get No Tracks Available on the PS3.
Photos work somewhat, but are slow.

A Little Blue Linux Box

The little blue Linux box seemed like a good candidate for a server. It now has a wireless adapter – ZyXEL G202 and is running Ubuntu 7.10.

After a little research, gMediaServer seemed like a good server to try. 0.12.0 is built into Ubuntu 7.10. The latest version 0.13.0 supports PS3 explicitly with a profile for PS3 that sets all the media types supported by the PS3. Not sure what else the server knows about the PS3.

When started with the following command line the server serves photos to the PS3 with no problems.

gmediaserver -ieth1 -v1 --file-types='jpg,mp3' --friendly-name='Blue-Diamond Media Server' /home/darrell/Pictures/Fantasy_1280 /home/darrell/Music/musicfiles

When running the gmediaserver only requires 1.6MB of memory. It serves photos with no problem and much faster than the Nero system, as observed from the PS3 system. Occasionally a “DLNA protocol error 2103″ appears, but this only appears once or twice when the server starts to display images.

Music files do not work. I get “No Tracks Available” on the PS3. Since neither server will serve music files to the PS3, the problem may be in the PS3. Update: Even the sanitized mp3 files would not show up at all in the PS3.

Twonky Media Server

Twonky Media 4.4.4 is very light weight – only about 4MB of ram total for the two server processes, and has a nice web interface for configuration.

It works very well for photos, and shows music files, but the files will not play. The PS3 tosses a “800100B network error” when you try to play a track, after a very long time of waiting.

Summary

Nero Media Home – Ultra 8 version. Photos work. Music doesn’t show any tracks and it tosses 80710092 protocol errors. This thing is 60MB of ram [ can you believe it??] in three processes, and it keeps crashing. So dead in the water as far as I’m concerned.
gMediaServer 0.13.0 on Ubuntu 7.10 – 1.6MB of ram. Works great for photos. But again music does not work. Shows no tracks. Occasionally I get 2013 protocol errors.
Twonky Media 4.4.4 – Very nice with web config and light weight in memory – two processes of about 4MB total of memory. Photos work just fine. Music tracks show in the list but trying to play a track gets an 8001000B protocol error.

BTW, all the mp3 music files play just fine from a data DVD on the PS3 and on WinAmp on PC. And on the Linux system too BTW.

So, I conclude that the PS3 V 2.17 is broken for media services for playing Music.

Anybody prove me wrong?

What media server are you using to play MP3 music on your PS3, and where did you get your MP3 file?
Do we know that the PS3 is picky about it’s MP3 files?
Why should it be pickier about the MP3 files over the net than it is from a DVD?

Update: This PS3 appears fairly picky about mp3 files, esp when reading files from a MediaServer and the protocol apparently does not recover from a bad file.

Ubuntu and Windows Play Nice, but Not Out of the Box

Darrell — Sat, 23 Feb 2008 23:08:20 +0000

Ubuntu and Windows can work together on a network of course, using winbind and samba. But Ubuntu doesn’t come that way and the configuration of Samba needs some work to make it happen.

I’ve got my systems all file sharing now, but it took some doing since I’m not a Samba expert.

It is a long and sorted tale taking about a week of posts on the Ubuntu forums with lots of missteps along the way, but I’ll cut to the final answer. If you really want to read the post in all it’s gory detail, go to the Very Strange Network Addresses post.

Caveat: I’m not a Linux expert and while I’ve used and coded IP applications for years, my commentary below may be flawed. If you see a mistake, let me know and I’ll be happy to correct it.

Start with Ping
After trying to fire up Samba and letting it go, it was a real mess. So when you do this it’s best to take it in stages.

Since modern Windows machines use file sharing over Tcp/Ip, we should get the machines talking with each other via Tcp first. I would have thought that Ubuntu would come Out of the Box [OOTB] that way, but it doesn’t.

I suggest you start by installing winbind and fixing up your /etc/nsswitch.conf file. Thanks to i_m_bobo for putting me on the right track here. My solution is similar, but not identical to his.

If you are new to installing packages, use System >> Administration >> Synaptic Package Manager and search for winbind. Set it to install and use apply. All new packages are installed this way.

I had to look around a bit to find out how to edit config files on Ubuntu. As you will quickly find, Ubuntu has a philosophy of a locked root account and uses sudo and gksudo commands to perform root command functions from your normal account. After using this a while, I quite like it.

Use sudo for command lines and gksudo for graphics commands. Look up these commands on the web for a full explanation, but here is how you edit your nsswitch.conf file:

darrell@squall-ubuntu:~$ darrell@squall-ubuntu:~$ cd /etc darrell@squall-ubuntu:/etc$ gksudo gedit nsswitch.conf darrell@squall-ubuntu:/etc$

The nsswitch.conf file I’m using looks like this:
darrell@squall-ubuntu:/etc$ more nsswitch.conf # /etc/nsswitch.conf # # Example configuration of GNU Name Service Switch functionality. # If you have the `glibc-doc-reference' and `info' packages installed, try: # `info libc "Name Service Switch"' for information about this file. passwd: compat group: compat shadow: compat hosts: files wins mdns4_minimal [NOTFOUND=return] dns mdns4 networks: files protocols: db files services: db files ethers: db files rpc: db files netgroup: nis darrell@squall-ubuntu:/etc$
I added “wins” after files and before dns so that local host names would be resolved by wins if possible before using dns. Without winbind and this change, the host names were being resolved externally by dns, which is bound to give bad addresses.

After this change, you should be able to ping the rest of the windows boxes on your network, whether they are getting their IP addresses via DHCP or they are set to static addresses. I have both types on my network. It’s a good idea to test the pings both ways, of course.

Install Samba and Friends
Now install the various Samba files. These are the ones that I have installed:

Click for a larger view.

Edit Your smb.conf File

I’m using WORKGROUP and not a domain solution.
Here are the changes that I’ve made:

wins support = no for do not make this machine a wins server.
Leave wins server commented out since we are not using wins in samba, but rather we are using winbind though dns.
dns proxy = yes looks like another good thing to do since we are using dns and not wins or netbios to resolve addresses.
Make sure that your name resolve order contains wins.

Authentication Settings

I would like to use security = user but my windows boxes refuse to connect to my Linux box with this setting. They prompt for a password, but apparently the Linux box does not accept the user / password. Not clear why. I have used encrypt passwords = false and no change.

Use System >> Administration >> Shared Folders to Create Shares

Once you have the machines connecting to each other, you can use System >> Administration >> Shared Folders to create shares. This program modifies smb.conf for you to add shares. Don’t mess with the Advanced tab tho since that may undo some of the above changes.

Apparently every time you make a change to the smb.conf file and save it, samba takes the new settings. No reboots required.

I hope this keeps some of you out of the weeds.

Update: Here’s the fix for security = user to work.

You need to add a password to the separate Samba password database. Use the following for each user and change security = share to security = user.

darrell@squall-ubuntu:/etc/samba$ sudo smbpasswd -a darrell [sudo] password for darrell: New SMB password: Retype new SMB password: darrell@squall-ubuntu:/etc/samba$

Enjoy
- windy

XML Signatures to Protect Settings Files

Darrell — Thu, 07 Feb 2008 00:48:03 +0000

When settings files are transferred as part of a program update, it may be interesting to assure that the files are not corrupted, or changed in such a way that the program is compromised. One can do this with XML Signatures.

Consider the following example:
Widgets Inc needs to create WidgetsSpecs.xml at their home office to update the WidgesFactory software at some of their many plants. In this example, WidgetsSpecs.xml is not a secret formula, but Widgets Inc wants to make sure that folks do not tamper with the settings along the way and create subversive or corrupted widgets. The following method will secure the WidgetsSpecs.xml file against tampering.

To security their specifications, Widgets Inc needs to create a private key, a self signed certificate, and then sign their WidgetsSpecs.xml files with the private key. Only the self signed certificate is passed to the application in the field and it is provided in a secure way so that it cannot be compromised. Once this is done then the software application can verify that the WidgetsSpecs.xml files are authentic.

Widgets Inc creates a Certificate Authority, using OpenSSL, and a private key of say, 2048 bits so that it is highly secure. 2048 bits of RSA key are the normal suggested strength for computing today. This process of creating a Private Key and Self Signed Certificate is outlined here and will not be covered in detail in this tutorial. The result is this is a private file called WidgetsInc.key which is the private key and will be protected in a very secure manner by Widgets Inc. and the WidgetsCert.crt which is the self-signed certificate which is public and need not be secured.
Once these are created, then the WIdgetsSpecs.xml file is created with some extra wrappers to contain the signature. There is an example below.
At the Widgets Inc company, in a secure place, the WidgetsSpecs.xml file is signed using the WidgetsInc.key file. This adds a digest or hash value to the xml file which can only be verified with the correct public key from the certificate.
When the WidgetsFactory application was built, the WidgetsCert.crt certificate was included with the application in a way so that it could not be compromised. If this crt file can be replaced by a hacker, then any signed XML file can be substituted. There are several ways to secure the crt file: (1) embed the file in the program itself rather than just placing the file on the disk where it can be changed. (2) In addition, embed a hash function of the crt file in the program so that the program can find out if the file has been tampered with.
Each time the WidgetsFactory application starts up, it reads the WidgetsSpec.xml file and verifies the signature using the valid crt file. If the CRT file does not pass the hash checks, or the signature does not pass, the WidgetFactory application refuses to make widgets.

Cautions and Caveats
Widgets Inc could decide to spend the money on a trusted certificate from a company like Verisign for this purpose. But these certificates cost money and they expire. When creating a Self-Signed certificate there are no costs and the expiration date can be set to 5 years, if desired.

The private key is protected at Widgets Inc and is not embedded or distributed with the Widget Factory application.

Example Files

These examples were taken in part from the XML Security Library site.

Template File including the Settings to be signed.

Signed XML File

Using xmlsec1 to Sign and Verify the XML files.
darrell@squall-ubuntu:~/code/xmlsigntest$ xmlsec1 --sign --privkey-pem ca.key --output WidgetsSpecs-signed.xml WidgetsSpecs-Template.xml darrell@squall-ubuntu:~/code/xmlsigntest$ xmlsec1 --verify --pubkey-cert-pem wwcdd.crt WidgetsSpecs-signed.xml OK SignedInfo References (ok/all): 1/1 Manifests References (ok/all): 0/0 darrell@squall-ubuntu:~/code/xmlsigntest$

To examine all the files in the example, download widgetsexample.zip.

Some observations:

This example was performed using Ubuntu 7.10, where xmlsec1 and the other software required is easily installed.
The software is easily ported to Windows and the libraries have been ported to Windows by Zlatkovic.
Enjoy,
ww

Puppy Linux on a Blue Diamond VIA C3

Darrell — Fri, 11 Jan 2008 01:25:20 +0000

Table of contents for blued

I recently purchased a 15″ PhotoFrame for what seemed like a great price of $250US. Of course now they have a WiFi version for $270. But I have a 17″ LCD monitor that wasn’t pulling it’s weight anymore, so I wondered how much trouble it would be to make a PhotoFrame from that.

Seemed like Linux was the way to go, so I looked around for a cheap, small format PC, since it didn’t make sense to run it with an Old Desktop machine. How Klunky is that?? Here is my final solution. I’m going to talk about what to watch for as you try to build one of these for yourself.

After a lot of looking around and being confused by all the options and lack of any support in this space, I found a company that had a clue and would talk to me about my application. The Linux Forums were no help, by the way. Several posts to several different forums got no response as if nobody had ever made Linux run on such devices.

This is the device I selected: Blue Diamond CV860A. I got the version with 800MHz but fewer Lan ports. I chose the CV860A – 1R80 since it was only $8 more than the slower model. Beware, this device uses “Low Profile” DIMMs. NewEgg had a 4GB flash card and a PC133 512MB DIMM for about $40 each, so I seemed all set.

Woops – Need to have Low Profile DIMMs

Until I found that the BlueD required Low Profile DIMMs.. But alas I lucked out and the PC133 was low profile and worked like a champ.
Here’s what happens with High Profile DIMMs – like the normal ones that you use in Desktops.

Can’t close the case. But low profile DIMMs work just fine:

Loading Linux?? From where?

The BD will support a 2.5″ IDE drive, but I wanted to run it from a CF card if possible. But how do you get it on there? Turns out the simplest way is to boot from CD-ROM and run an installer. Well, the BD supports a USB CD-ROM, so no problem. Here is the system all hooked up and ready to install:

Ubuntu or not to Ubuntu?

I had installed Ubuntu on a desktop machine and a HD. No problem, well almost no problem. The system set the screen to 800×600 and the installer dialogs were assuming a larger screen, so that was a hassle, but after that the system installed just fine. On the desktop….

But on the blue diamond, it failed to start. It got all the way to the GNOME display manager and then hung. The system on the CDROM would not start. I’ll post this on the Ubuntu forums and see if anybody has any ideas.

Update: 14 Jan The problem with the Ubuntu install was insufficient RAM. The Ubuntu CD requires 384MB RAM for the install. Once I tried it with 512MB of RAM it worked fine. Ubuntu 7.10 installed without problems on a 38GB IDE 2 1/2″ HD connected to the Blue-Diamond. The BD is supplied with a cable and bracket to hold a standard Laptop form-factor HD. While Puppy works fine, as you will see from the later articles, Ubuntu is a better match to take advantage of the HD and RAM space of the larger system. On the BD, Ubuntu correctly sized the screen for 1280×1024x24, unlike the previous install on a Compaq Presario.

Continue with the series to read about Puppy Linux installation.

Next in series

Booting Puppy on the Blue Diamond

Darrell — Thu, 10 Jan 2008 23:34:29 +0000

Table of contents for blued

I chose Puppy Linux 3.01 and burned a CDROM. I’ve got Nero 8, so that part was a snap.

When I booted the CDROM, I got a much better result than for Ubuntu:

and then it popped up this:

Looked like it was rolling. Next was the Video wizard:

And then after choosing XORG I got this:

The BD has a limited Video chip. It supports 1600×1280x16, but I needed 24bit color at least for photos. But it turns out that this works fine. And then you get a friendly Test Dialog:

and the test was fine… Two CTRL/ALT/BACKSPACEs to get back to the wizard and

All was well.

Now the next part is installing Puppy on the CF flash card.

Previous in series Next in series

Installing Puppy on the Blue Diamond

Darrell — Thu, 10 Jan 2008 22:37:45 +0000

Table of contents for blued

The next problem is installing Puppy on the CF card.

After looking around a little, the installer was right there in the Setup menu. And the choice was pretty clear:

Woops Can’t Find the Files.

You would think that the correct choice would be CD in this case, but this fails with an error saying that it can’t find the files.

The problem is that the CDROM is not mounted as the system boots from it. So all you have to do is click the button on the right of the CDROM to mount the device so it looks like the following. At this point the CD choice will work correctly. Perhaps the correct way to do this is to mount the CD before running the installer to avoid the error completely.

When it finds the files, you get the following dialog.

Do nothing was the right choice here, but for a while I got confused and tried something else. But I backed out before actually writing anything funny on the CF card.

Finally the installation started:

And then I booted it and it ran just fine:

After booting the system and making some changes, I triggered a reboot and it asked if I wanted to save the settings. The size it suggested was 512MB, which seemed pretty large, but apparently what it wants to do is save a squashed file system. Later on I reinstalled Puppy on a 512MB card and reduced the size of the saved file to 64MB and it worked just fine. I’m not clear at this point what is actually saved in that file system. But for the photo-frame application, it works just fine with a 64MB saved size.

Once the settings are saved, the state of the system is written to that file every time you shutdown the system. I’m not even clear on how to cause another save file to be created. But every change you make is now saved when you shut down the system gracefully. Here are some things you will want to change for a Photo Frame application:

Disable the ScreenSaver and Energy saving in Menu>>Desktop>>Gxset.
Change the background from the default Woof Woof using Menu>>Desktop>>PuppyDesktopBackground Image
Run GTKSee graphics application, which is what I use for the slide show, and set the background, slide duration and other settings. Then exit the program cleanly to save the settings.
And any date and time zone settings.

When you’ve made all your changes, just use Menu>>Shutdown>>Restart to save your settings. This time you will not be asked about a save file and all your settings will be the same after reboot.

So there it is. A Blue-Diamond running Puppy Linux to make a photo frame from a 17″ LCD Monitor. I’ll look into what it takes to run other programs or make changes to the Slide Show program in the future. But for now, problem solved.

The final CF Flash drive usage was about 165MB including the 64MB save file, so the Puppy Linux system only takes about 100MB of flash space. Amazing for what you get.

Previous in series

Problems Installing Video on Linux

Darrell — Fri, 17 Nov 2006 22:28:08 +0000

I’ve installed video and dvd support as discussed in this article in PCTech101.
At this point I’ve installed RPMs from the OpenSuse repository. http://software.opensuse.org/download/repositories/KDE:/KDE3/SUSE_Linux_10.1/

I had to install these rpms manually. YAST refresh of the site would never complete and it would then produce errors that the available RPM could not be found on the site. I downloaded them from the site and then installed them using rpm -Uhv.

These are the components that I have installed in the KDE mm suite. [click for larger image].

When I use any of the video players: Totum, Kaffein, Xine – the audio plays, but there is no video. Is there something else that I need? I get a black screen. No errors. The configuration for Kaffein indicated no errors.

Thanks,
ww

CVS with KDevelop and Cervista

Darrell — Fri, 10 Nov 2006 04:06:18 +0000

[Updated 10 Nov 2006]

I’ve just started learning how to use CVS on Linux with KDevelop, and there are enough gotchas that I thought it might be interesting to share what I’ve learned. I have used lots of source control in the past, but never CVS. RCS [ages ago], QVCS [windows front end for rcs], and VSS with Visual Studio.

Serious coding on Linux means that I need source control, so I had a look at what’s available. CVS seems to be the long-time standard for open-source projects. But SubVersion is new on the scene, relatively, so I took a look. But after a little study of a few doc pages, including this one, it looks like SubVersion has the nasty habit of suggesting, if not requiring, that your directory tree look in a special way for SubVersion use. That, and the fact that the SubVersion GUI client for Linux looks like it is in its pretty early days, has lead me to choose CVS and Cervisia. Any Open Source code base with a version above 1.5 has got to have seen a fair amount of use.

So, here are my experiences with SUSE 10.1, KDevelop 3.3.5, Cervisia 2.4.5. I chose to build a local repository that is file based rather than a server based repository.

Install Cervisia and CVS. All on the original distro.
Figure out how to create a CVS repository. Turns out that Cervisia Repository >> Create… works. Note that CVSROOT will be created inside of that folder. So don’t end the path in CVSROOT.
At this point you cannot use File>> Open Sandbox. To open a sandbox you must have done an Import and a Checkout..
I was nervous, so I backed up my KDevelop project folder which already was in play. Then use Repository >> Import.
The parameters to Import are a little tricky. The Respository path does not end in CVSROOT so it is
/home/darrell/Development/CVS in my case, not
/home/darrell/Development/CVS/CVSROOT.
The module is a name that will be applied to the entire import, so a lowercase project name without spaces and no extension is appropriate. I used khexwin for example. This should be the same lower case name as the project folder.
The working folder for Import is the project folder. For Checkout it will be something different so beware.
I just used wwc and release for the tags and left the rest blank for now.
You should see success when you import your project.
At this point I would suggest you rename your project folder to myprojectORIG and then
do a Repository >> Checkout…
The Repository folder is the same as before
The module is the same as before
The working folder will be the containing folder, not the project folder. So if your project path is
/home/user/Development/project then you will want to use
/home/user/Development/ here.
The project will come from the module name. If you want to change the folder, then you can put a new name in Check out as:. ~~But KDevelop will freak out if you do that, at least I don’t know yet how to recover from a change in the project path~~. The project will need some adjustment to build if the path is changed.
Don’t forget to check Recursive checkout.
Warning: KDevelop gets very upset if you change the path of a project. There is probably a way to recover after you move a project, but I don’t know what it is. So you will have to check out your project to exactly the same path as you created the project, or all hell will break lose and the project will not build.
After the checkout you should be able to use KDevelop to open the project and build it. All should be well. If the path of the project has changed, use Build >> Clean Project, followed by Build >> Run Automake & Friends and Build >> Run Config.
Cervisia can now do a File >> Open Sandbox… since there are CVS subfolders in the checked out tree. I have not turned CVS support on in KDevelop. It looks like you can check files in/out in the AutoMake panel on the right. I haven’t seen other places that light up with checkin/out menu items.
I just use Cervisia to commit files. If you are working alone like I am, you won’t have to merge or update based on other folks work.

I had some issues:

~~As I mentioned, KDevelop has problems if you change the path to the project. I’ve reported a BUG and posted a forum topic to get clarification about how to get a project to work after it is moved.~~
When I tried to check in the files after I changed them, I was getting errors about a sticky tag – release – that was used to Import the project. I took a branch to solve this. Use Advanced >> Tag/Branch… But I think a better way would be to just add a new tag rather than a branch. At this point I have merged the branch into the main branch so things look a lot more sane.
Be sure and set the Recursive setting when doing a commit so you can just checkin from the top of the project tree.
Use Update [?] often from the tool bar to see what you have changed. The coloration of files is pretty obvious and very helpful.

Looks like CVS / Cervisia is working for me now.

Good luck,
ww

HexDisplayWidget begins to work

Darrell — Sun, 05 Nov 2006 19:57:07 +0000

I’m working on a Hex Display Widget under KDevelop.

Here’s a first version that begins to work. The options in the view menu work to change the display grouping.

ww

KHexWin test program

Configure VNC on SUSE 10.1

Darrell — Fri, 03 Nov 2006 17:29:12 +0000

VNC, or remote administration, allows a remote desktop service between windows and linux. If you are like me you have a mixed network, and it is nice to be able to reach the linux box from a windows box. One my major interests is to be able to run a Kdevelop app from my windows laptop while I’m watching TV.

To accomplish this, make sure the software is installed using YAST Software Management:

Perhaps not all of these are required.

Under Network Services in YAST, click on Remote Administration, and enable it. Also choose whether you want to use a firewall. I have a router between my local network and the Cable Modem so I do not use the firewall.

By default only one VNC service, host:1, is enabled. host:1 is the string that most VNC viewers will want to connect to your host. The default service gives you a 1024×768 window. This wasn’t large enough to allow me to run KDevelop, which is my primary reason for using VNC.

If you are using Network Manager, like I am, to configure the other VNC services, use YAST and In Network Services group click on Network Services(xinetd)

At this point you see a list of services. In the list are vnc1, vnc2 and vnc3 services. From a VNC client, on windows or Linux, you use host:1 host:2 host:3 in VNC Viewer to get to these services.

Just had to turn them on in this YAST panel. I don’t know where the port definitions come from. They aren’t in /etc/services and they aren’t here to be edited. So apparently some other file relates vnc1-3 to ports 5901-03 to vnc1-3.

You can change the geometry as well as other options in this panel. I edited vnc 3 to use geometry 1600×1050 to match my wide screen monitor and voila… Back on my 1920×1024 laptop I have a have a 1600×1050 window to my SUSE machine.

It may be best to turn off any wallpaper desktop background when using VNC to improve performance. For a windows client, I use the open source, tightVNC. There is also a free version of the client at Real VNC

enjoy,
ww
.

Hellraiser Box for Linux

Darrell — Sun, 29 Oct 2006 00:42:12 +0000

This is a port of the Hellraiser box to Qt / KDE / Kdevelop on SUSE 10.1.

To run the program, use fileroller to unpack the file in a directory and then run the program. The image files must be in the same directory as the program. The instructions are essentially the same as for the Windows Version.

Hellraiser Lament Box Program

The sources are provided here. They are built using KDevelop 3.3.5 and Qt.
Hellraiser Lament Box Sources
There are links in the debug and optimize src directories that point back to the images in the src directory. These may have to be remade after you unpack the archive since links are not relative.

Enjoy.

Apache mod_rewrite solved

Darrell — Mon, 16 Oct 2006 18:00:15 +0000

I found out some of the great mystery of mod_rewrite. Here’s an example that works. Turns out there are errors in the documentation, of course, and this article may help avoid the pitfalls that I have discovered.

Here’s an .htaccess file that works for the example that is included here:
# Turn the re-write engine on so it rewrites URLs RewriteEngine On # not needed since we are sending URLs back to this same directory #RewriteBase /test-rewrite/ # if the URL does not point to a file RewriteCond %{REQUEST_FILENAME} !-f #and if the URL does not point to a subfolder RewriteCond %{REQUEST_FILENAME} !-d # be tolerant of trailing slashes or not RewriteRule ^products$ products/ [NC] # and if there is no product specified, list them all RewriteRule ^products/$ /content.php?action=product_list [NC] #products can have a code and a subcode RewriteRule ^products/([0-9]+)/([0-9]+)$ products/$1/$2/ [NC] RewriteRule ^products/([0-9]+)/([0-9]+)/$ /content.php?code=$1&subcode=$2 [NC] #or if we specify the code only, then forget the subcodes and list all the products in that code RewriteRule ^products/([0-9]+)$ products/$1/ [NC] RewriteRule ^products/([0-9]+)/$ /content.php?code=$1 [NC]

I discovered a bunch of rules that are not stated explicitly:

In RewriteRule, no slashes in patters after the ^
Slash at the beginning of the target after the pattern.
In a RewriteRule to make slashes optional, do not use [R] as the doc says. It doesn’t work.

The example that’s included here does work. It’s been tested. Unlike most of the rest of the examples out there. Sigh…

Apache Mod_Rewrite Example

Apache Configuration

Darrell — Wed, 02 Aug 2006 16:58:38 +0000

~~Anticipating your request, here is the entire apache configuration tree.~~
This one is broken.
Broken Apache Configuration

[SOLVED]
I got this working. Here’s an example configuration tree for SUSE 10.1, Apache 2 and php. Don’t just dump these into your system. I suggest doing a diff – kdiff3 is a good tool – and understand what the differences are before you apply them to your system.

I found that the OUT OF THE BOX experience with Apache was not good. The config for apache is designed to be basically turned off. So you need to make changes to enable anything. You do not need to “reinstall the whole thing” as some folks suggest to get it working. That won’t do anything except increase your level of frustration.

Here’s a working config – at least it works for me.

Working Apache, and php5 config files.

enjoy. I’m not an expert, but if you contact me, I may be able to assist.

Gears OpenGL for Linux

Darrell — Sat, 15 Jul 2006 02:23:24 +0000

A part of the Geodesic program for WIN MFC and OpenGL has been ported to Linux / KDevelop / Qt / OpenGL. It creates and displays a gear object and there are many parameters to change the shape of the gear. The program will be able to export OBJ or POV files of the shape of the gear. It can serve as a complex tutorial of using OpenGL and many other features of
the KDE and Qt frameworks.

Ring gears can be created as well:

There are many options to change the view of the object:

There are many properties to adjust the shape of the gear:

The display of the gear can be rotated, moved and resized using the mouse. Clicking down and dragging the mouse in the window rotates the object. Using SHIFT moves the object and using CTRL resizes the object. In addition, using animation from the View menu allows the object to be spun with the mouse in the windows. The object can be displayed using OpenGL display lists or using direct OpenGL commands. Of course using Display Lists, which are pre-compiled OpenGL commands, are much faster.

Download the KDevelop project and source here:

Download 2006_07_14 version of GearsOGL

Qt3 Validators Work

Darrell — Fri, 07 Jul 2006 20:43:47 +0000

There is a properties dialog with a large number of input fields as follows:

Here is the header definition for the data.

// Way too much junk to pass back and forth. // So let's make it all a class and then we can use class copy to move this stuff // around. class CGearParams { public: double m_dBaseWidth; // width at base of tooth. This sets diameter of gear. double m_dGearThick; // thickness of gear outer ring and hub double m_dSpokeThick; // thickness of spokes [or disk of 1 spoke] double m_dHubRadius; // radius of hub around shaft double m_dHubThick; // thickness of the hub double m_dRingWidth; // width of outer ring of gear double m_dShaftRadius; // radius of shaft int m_nHeightPct; // Height of tooth as percent of base int m_nTopPct; // Top of tooth as percent of base int m_nTeeth; // number of teeth int m_nRadialSecPerTooth; // radial sections, per tooth bool m_bSpiral; // spiral type gear? bool m_bSpiralLeft; // Reverse the spiral int m_nSpiralSections; // how many sections along height for spiral int m_nSpiralAngle; // spiral angle - 0 = no spiral, 89degrees very shallow int m_nSpokes; // how many spokes int m_nSpokesPct; // spokes radial percent // each spoke takes up this much of its angle bool m_bRingGear; // Create a ring gear with teeth inside }; // // IITEM - integer item // DITEM - double item // CITEM - checkbox item // IITEM( widget, member, min, max, default, name ) #define GEAR_OPTIONS_ITEMS \ DITEM( gearThickBox, m_dGearThick,0.2 , 1.0, 0.5 , "Gear thickness" ) \ DITEM( spokeThickBox, m_dSpokeThick, .01,1.0 ,0.125 , "Spoke thickness" )\ DITEM( shaftRadiusBox, m_dShaftRadius, .01,1.0 ,0.125 , "Shaft radius" )\ DITEM( hubRadiusBox, m_dHubRadius, .01,1.0 ,0.25 , "Hub radius" )\ DITEM( hubThickBox, m_dHubThick, .01,1.0 ,0.35 , "Hub thickness" )\ DITEM( outerRingWidthBox, m_dRingWidth, .01,1.0 ,0.25 , "Outer ring width" )\ CITEM( reverseCheck, m_bSpiralLeft,0 ,0 ,false , "Spiral to the left" )\ IITEM( numTeethBox, m_nTeeth, 10, 60, 20, "Number of teeth" )\ CITEM( spiralCheck, m_bSpiral,0 ,0 , false, "Spiral gear" )\ IITEM( spokesBox, m_nSpokes,1 ,20 ,1 , "Number of spokes" )\ IITEM( radialSectionsBox, m_nRadialSecPerTooth,1 , 10,4 , "Radial sections per tooth" )\ IITEM( spiralDegreesBox, m_nSpiralAngle, 0, 89,45 , "Spiral angle in degrees" )\ IITEM( spokesWidthPctBox, m_nSpokesPct,10 , 500, 10, "Spokes width %" )\ IITEM( spiralSectionsBox, m_nSpiralSections,1 ,20 , 4, "Spiral sections" )\ CITEM( ringGearCheck, m_bRingGear, 0, 0, false, "Make ring gear" )\ DITEM( baseWidthBox, m_dBaseWidth, .01,1.0 ,.25 , "Base width of tooth" )\ IITEM( heightPctBox, m_nHeightPct, 50,150 ,100 , "Height % of tooth" )\ IITEM( topWidthPctBox, m_nTopPct, 20, 100, 50, "Top width % of tooth" )
The code to set the validators looks like this:
// // define the macros to set the validators and the default values // for the GUI items #define IITEM( widget, member, min, max, default, name ) \ { widget->setValidator( new QIntValidator( min, max, this, name ) ); \ widget->setInputMask( QString("0000") ); \ QString str; str.setNum(default); widget->setText( str ); m_params.member = default; } // #define DITEM( widget, member, min, max, default, name ) \ { widget->setValidator( new QDoubleValidator( min, max, 3, this, name ) ); \ /*widget->setInputMask( QString("0.000") ); */ \ QString str; str.setNum(default); widget->setText( str ); m_params.member = default;} // #define CITEM( widget, member, min, max, default, name ) \ widget->setChecked( default ); m_params.member = default; // void GearOptionsDialog::setOptionsValidate() { GEAR_OPTIONS_ITEMS; }

The code to check the validators looks like this:
// Check the validation and report any errors with an intelligent message bool GearOptionsDialog::validateItem( QLineEdit* widget, const char* name, QString& mins, QString& maxs ) { bool bOk = widget->hasAcceptableInput(); QString errstr; if ( !bOk ) { errstr = name; errstr.append( " is out of range:\n"); errstr.append( mins ); errstr.append( " - " ); errstr.append( maxs ); KMessageBox::information( parentWidget(), errstr, QString("Invalid Input") ); } return bOk; } // #undef IITEM #undef DITEM #undef CITEM // define the macros to set the GUI items from the members #define IITEM( widget, member, min, max, default, name ) \ /* bErrors |= !widget->hasAcceptableInput(); */ \ {QString mins;mins.setNum(min); QString maxs; maxs.setNum(max); \ if ( !validateItem( widget, name, mins, maxs ) ) { bErrors=true; goto cleanup; } } // #define DITEM( widget, member, min, max, default, name ) \ /* bErrors |= !widget->hasAcceptableInput(); */ \ {QString mins;mins.setNum(min); QString maxs; maxs.setNum(max); \ if ( !validateItem( widget, name, mins, maxs ) ) { bErrors=true; goto cleanup; } } // #define CITEM( widget, member, min, max, default, name ) // bool GearOptionsDialog::validateData() { bool bErrors = false; GEAR_OPTIONS_ITEMS; cleanup: return !bErrors; }

hasAcceptableInput() works correctly now that the validator objects are “new” rather than being stack items. A misunderstanding about whether they were copied into the QLineEdit object. This code provides an intelligent popup box to fix any errors. The integer objects use input masks, but the double input mask didn’t allow valid input to be entered.

The validators appear to be working now. A future post will contain the valid project.

Linux Progress

Darrell — Fri, 07 Jul 2006 06:35:12 +0000

Work continues with SUSE 10.1 Linux both in configuring the system and in learning to use the KDevelop IDE for C++ programming. There are several interesting items to mention:

Overlay planes
Wacom Tablets
KDevelop glitches
.ICEAuthority lockouts
GAIM messenger
SKYPE
CD backups

An OpenGL application that displays on the Overlay plane did not work with built-in drivers. After building and installing the latest NVIDIA drivers for the GF4200 graphics card, still no joy for overlay support. Possibly some parameter needs to be changed, but this needs to be done essentially manually. SAX2, the X server administrator, works pretty well, and has a test mode, but it doesn’t have a plug-in panel that allows drives to supply dialogs to configure graphics cards. But 3D accelleration works and the next programming project will be to port over Gears from windows to display and export gear objects. The KDE mascot is a gear. Installing the NVIDIA drivers apparently deleted the MESA OpenGL libraries, so project wouldn’t build without redirecting which library was used for OpenGL.
A Wacom Graphire 2 graphics tablet worked well when attached to the USB ports. But GIMP doesn’t show pressure sensitivity. Further work on this problem is required.
KDevelop continues to show some glitches, but they are managable. Few crashes, but occasionally GDB [debugger] fails to work. Closing and reopening the project appears to fix this problem. Learning to configure AutoMake for specific include files and libraires has been part of this learning experience as well.
Occasionally .ICEAuthority, a file in a users home directory, becomes either locked or owned by root, which causes headaches for the KDevelop IDE and for launching applications in general. The fix is to use a root file browser to delete any lock files [.ICEAuthority_l and .ICEAuthority_c] and to change the owner of the file back to the current user from root/root. It is not clear why this owner change occurs, but it may be due to running some applications as root to maintain the system.
GAIM instant text messaging works fine with both MSN and Trillian. GAIM will eventually have audio chat built in but not yet. SIMP secure chat doesn’t work with GAIM of course, and no response to my email to the Secway folks about that issue. Sending files from GAIM to Trillian doesn’t work, but it didn’t work with MSN Messenger to Trillian either. Sending files between GAIM and MSN works just fine.
SKYPE audio phone/chat looks like a first class product on Linux. The SKYPE folks are really serious about Linux support. I’ve spent several hours on SKYPE to windows users without problems. Audio is high quality and setup was trivial. Performance is excellent even on my antiquated 400MHz Dual PII system.
Backup of my files to CD should be a priority. If the whole thing goes down the tubes now, significant work will be lost in the KDevelop projects and book marks. Firefox has an excellent bookmark export facility so that will be no problem. The bookmarks are exported as a single HTML file so they can be imported into any browser by launching the file and manually saving them.
After solving the localhost permissions problem with MySQL users permissions, Apache-MySQL-php-WordPress is all running just fine. And the move of the web base files to the /home partition was no problem after adding the Directory option to the Apache configuration.
A 9GB disk is not big enough for a Linux system with any sort of development and a webserver. Installing the NVIDIA drivers required installing the kernel headers and libraries, which ate up another 200 MB of my root partition. The root partition is now 80% full. In a few months I’ll have to rebuild this whole system on a larger disk to keep using it.

Pressing on and loving every minute of it.
d

KDevelop project stuck

Darrell — Mon, 03 Jul 2006 22:30:16 +0000

SUSE 10.1, KDevelop 3.3.3, KDE 3.5.1

RESOLVED: A reply in a forum indicated that there is a known bug with the default configuration that leaves a config.status around that sabotages the entire project. The recovery is to delete all the config.status files in the project and then avoid use of the default configuration.

Developed the program under the debug configuration and then tried to switch to default or optimized configuration. Tried many combinations to try to get it working again, but all BUILD options give errors ending with makedistclean.

such as :
cd '/home/darrell/Development/siteprobe' && WANT_AUTOCONF_2_5="1" WANT_AUTOMAKE_1_6="1" gmake -f Makefile.cvs && cd '/home/darrell/Development/siteprobe/optimized' && CXXFLAGS="-O2 -g0" "/home/darrell/Development/siteprobe/configure" && cd '/home/darrell/Development/siteprobe/optimized' && WANT_AUTOCONF_2_5="1" WANT_AUTOMAKE_1_6="1" gmake This Makefile is only for the CVS repository This will be deleted before making the distribution *** automake (GNU automake) 1.9.6 found. *** Creating acinclude.m4 *** Creating list of subdirectories cd . && gmake -f admin/Makefile.common subdirs *** Creating configure.files *** Creating configure.in cd . && gmake -f admin/Makefile.common configure.in ; *** Creating aclocal.m4 *** Creating configure *** Creating config.h template *** Creating Makefile templates *** Postprocessing Makefile templates *** Creating date/time stamp *** Finished Don't forget to run ./configure If you haven't done so in a while, run ./configure --help checking build system type... i686-pc-linux-gnu checking host system type... i686-pc-linux-gnu checking target system type... i686-pc-linux-gnu checking for a BSD-compatible install... /usr/bin/install -c checking for -p flag to install... yes checking whether build environment is sane... yes checking for gawk... gawk checking whether make sets $(MAKE)... yes configure: error: source directory already configured; run "make distclean" there first *** Exited with status: 1 ***

So, apparently the project configuraiton has been corrupted and don’t know the way back to sanity.
Archive of the entire project here as a tar.gz
Siteprobe archive

thanks,
ww

php5 and MySQL 5 – Access Denied

Darrell — Mon, 03 Jul 2006 20:58:09 +0000

When setting up a test website using Apache 2, php5 and mysql5, the connection to the database failed with AccessDenied. If the password were removed, then the failure looked different. The resolution to the problem is to include localhost explicitly in the host list and set database privileges for localhost. The default is @% which is wildcard, but apparently wildcard does not include localhost.
For a complete description, screen shots and the whole story:

Here is the MySQL Admin screen for users.

To add a host, select the user and right click. Then choose Add Host and choose localhost.
Then set the privs for that connection. Select localhost and go to the privileges tab.

Select all the privileges in the right list and click the (< ) button to move the privs to the allowed list.

Now, php5 and mysql5 will play nice together under apache. So for example you can install WordPress.

cheers,
dd