About RTMP

RTMP is not rtsp (open, RFC2326) and not mms (Microsoft). It uses (in its TCP variant) a persistent binary data connection to enable a player application to deliver constant quality. Most (all) players are implemented in Flash. Besides the plain TCP versions, there are variants called RTMPT (which allows tunneling) and RTMPS (for https transmissions).

A more detailed overview explanation is available over at Wikipedia. According to the whitepaper, rtmp uses port 1935 (TCP), 443 (HTTP/SSL) and 80 (plain HTTP).

Development on deciphering the protocol is advancing. The protocol itself has been mostly reverse engineered by the Red5 project and others, making it available for free or open-source media server implementations.

Recording RTMP or downloading it

Most sites use a SWF/Flash player object embedded into a webpage to playback video streams delivered via rtmp. Furthermore, many player implementations seem to access a configuration xml file (instead of the video file URL directly) that, besides other stuff, stores the actual rtmp://path/to/flv/file path in it.At the moment there are only shaky solutions available to download data that is served via rtmp://. This is why so many content owners and video portals handling “watch-only” video (like hulu.com, BBC iPlayer, Mogulus) are using it in the first place.

The ingenious Mplayer, sadly, is still not processing it, neither for playback, nor for recording by using the “-dumpstream” switch. Write a comment if you know that this has changed!

Solutions that seem to support rtmp are: xbmc (linux, playback), vlc (multi-platform, serve in rtmp) ,  gnash project (linux, playback via the libamf library)  and rumours go that there’s a ffmpeg rtmp and libCurl is working on it. If you are on windows, orbit downloader seems to download rtmp.

And of course, as everything rtmp does travels over your network connection, you can dump/save everything via libpcap if you manage to get the binary dump right (I can’t). For the knowleadgable it might be helpful to have a look at Braden’s libpcap program.

Tell me if hotmail, gmx, or one of the other players in the email biz made a good interface for online email lately. So far I’ve only seen 1990s stuff out there. Szzzzz.

One of the limitations of domains secured with SSL certificates is that a certificate needs to be hardcoded to be only valid for a single subdomain + top-level-domain combination, like www.example.com. If your sitestructure requires you to setup subdomains like ftp.example.com your (purchased) SSL certificate won’t be valid for this subdomain. Every browser will complain, and for example Firefox 3 will completely hide the site until a good amount of clicking and accepting is done.

Anyway, this problem can be solved with what is called a wildcard certificate. On the creation of a SSL certificate, more precisely on the Certificate Signing Request step, you are asked to enter certain key essentials about the organisation asking for a request, among these details is the “Common Name (eg, YOUR name) []:” - which is the domain the certificate will be used for.

Under www. only operation you would enter “www.example.com” here. To enable a so-called wildcard certificate, you might assume you would enter “example.com”, but actually you need to enter “*.example.com” (a wildcard placeholder). This actually breaks standards but does magically work, furthermore even certification authorities (CAs) require this non-standard notation.

The downside of wildcard certificates is that CAs already charge for standard certificates and they want you to pay even more for wildcard certs! Time to propell development of the free CACert.org CA alternative and enable proper security for all of us without silly trouble.

As you might know you can also self-sign certificates, but this will also pop-up browser warnings and is thus not ready for production environments.

Hint: While debugging your SSL installation, this trick might help you debug your SSL.

The story was that the main drive C:\ with Windows XP Pro on it had a few defects and required a fresh XP install. Like certainly many systems these days, the box had another drive installed as D:\, which was a 400+ GB disk, used as a data grave. To prepare for the fresh install of the operating system, I bootet into the SystemRescueCd Linux LiveCD and moved all remaining non-OS related data from C:\ to D:\. After that I started the XP Pro Systembuilder install and about 45 minutes later was greeted with the green hills of the XP desktop.

The Adventure

So far so good, but then came the shock: Using the Explorer, navigating around the system, I noticed that the second drive, the giant data harddrive D:\, had disappeared. Gone. The system had identified the second drive, yes, but had somehow labeled it as unusable or ureadable. A look into the properties told me that the drive was “Not formatted”!, and is declared as being RAW. Or course I denied the question, if XP should now format the drive for me!

In a state of nausea and denial, I booted back into the SystemRescueCd, as before - under linux - the harddrive had shown up okay. Finally the relief: under Linux/ntfs-3g every bit of data was neatly in its place. Puh! But why doesn’t Windows recognize the drive properly?

My first guess was that the read-write enabled ntfs-3g, which I used to move the data, has somehow currupted the drive. Like most users that use ntfs-3g as a black box (non-ntfs-3g-developers), I am a bit suspicious if the write support is really that good. Wouldn’t Microsoft secretly bug NTFS and thus blast third-party efforts to access NTFS? But after reading this post- NTFS volume does not work in windows vista- I do think ntfs-3g is a mature application! Most of the drive identification under XP (is it really RAW?) is done through the Master Boot Record (MBR), and I learned that ntfs-3g operates on a completely different layer, in fact it has no access to the MBR at all!

So two options remained:

1. A virus or a hardware defect corrupted the MBR, thus rendering the drive inaccessible for XP (only because it does not identify it correctly) while ntfs-3g, which ignores the MBR, is completely happy with the drive.
2. The MBR, the format specification of the drive, whatever - is simply not readable for XP.

The Solution

Knowing that the drive’s data was 98% safe, as Linux/ ntfs-3g could properly access it, I  continued bringing the fresh XP install up to speed. Changed the graphics driver, network setup, the usual chore. Finally the install of XP Service Pack 3. A reboot. And surprise! Drive D:\ is back online!

But why? A physical defect that appears and disappears? Voodoo? Actually, no.

While installing XP Pro, I had time to muse about how old this XP install CD is. 2002! Compare this with the average update cycle of Ubuntu, Linux or even XP today! Remembering this notion while thinking about possible causes for this mysteriously diappeared and re-appeared harddrive gave me a bit of inspiration. So my best guess is that the old Windows XP has buggy or at least not full support for contemporary large drives. You won’t notice this with a patched SP2 and regularly updated system, but you will if you ever need to return to an older version of XP.

So all those desperate folks out there finally scrubbing their valued data after a drive appeared as RAW after a fresh XP install, try updating your system first! For example:

• Second hard drive goes from NTFS to RAW after XP install
• “Raw” hard drive after Win XP setup - File and Hard Drive Recovery
• Spare drive is now “RAW” after XPPro Clean Install
• After Windows Re-Install - Whole Harddrive Dissappeared!
• external hard drive erased when plugged in after a fresh xp install

And tell me afterwards if this worked for you or not.

Of course, please take this post with a grain of salt, don’t blame me if you loose data, all actions on your own risk! After all, there is a lot of guesswork in this post and I just documented what worked for me. There’s no guarantee that this will work for you.

The problem seemed be that I had no version of Wx on my system and the compile routine from Alien::wxwidgets failed as well, don’t know why. So I decided to first get a working version of Wx onto my system via apt. Follow this guide to get a version 2.8 on your system.

Running cpan again, answer the question to compile WxWidgets with “No” as it is already on your system. But soon you will find out that cpan has even more to complain: it can’t find wx-config. This is because we did not install the -dev packages. Do so by:

apt-get install libwxbase2.8-dev libwxgtk2.8-dev

Now the WxWidgets toolset is complete and you can install Wx from cpan. If it complains that it still has no luck, try one of the older version. In my case I had luck with Wx 0.83:

cpan> install MBARBON/Wx-0.83.tar.gz

That’s it.

1. Open the console/a terminal by pressing CTRL + ALT + T

2. Paste the code into the console window just beside $home>

3. Paste below, then ENTER:

wget -q http://wine.budgetdedicated.com/apt/387EE263.gpg -O- | sudo apt-key add -

4. Paste below, then ENTER:

sudo wget http://wine.budgetdedicated.com/apt/sources.list.d/etch.list -O /etc/apt/sources.list.d/winehq.list

5. Paste below, then ENTER

sudo apt-get update

6. Paste this, then ENTER

sudo apt-get install wine

7. Restart the EeePC!!
8. Paste below, then ENTER

winecfg

Just check the outputted info and see if everything seems okay. No action should be required.

Starting a program can be done by right-clicking on the .exe and choosing “Open with…”  > “wine”, just a double-click, or from the console with wine /<path>/<to>/<file>.exe
wine install explanation:

3. Fetches the hash key to correctly identify the wine apt package

4. Fetches the information about the location of the wine packages and adds them to the ../apt/sources lists

5. Rebuilds the apt-get “available packages” list now including wine

6. Tells apt-get to install wine

8. Is a utility that dies a basic init and tells you how wine is currently configurated.

Note on “speicherzugriffsfehler”:

If you get a strange “Speicherzugriffsfehler” when trying to execute win32 binaries using wine after the above steps, this is a hint that you skipped Step 7: “Restart your EeePC”! Wine is not yet fully installed, reboot and refresh your system setup - and the error should be gone!

SELECT table1.id, table1.fieldA, table1.fieldB, table2.id, table2.fieldC, table2.fieldD
FROM `table1`,`table2`
WHERE table1.fieldA = ‘test’
AND table1.id = table2.fieldD;

Here we try select’ing in a join’ed manner from two tables and merge the results into one resultsset. But MySQL will throw an error, as both tables have a column named “id”.

The solution to this problem is to tell MySQL an alias name for one of the identically named columns so it can differentiate between them in the results set. In our example here, we tell MySQL by using the “AS” constructor to use the obvious name “table1.id” for the “id”-field from table “table1″, while it should use the plain “id” for the “id”-field from “table2″:

SELECT table1.id AS “table1.id”, table1.fieldA, table1.fieldB, table2.id, table2.fieldC, table2.fieldD
FROM `table1`,`table2`
WHERE table1.fieldA = ‘test’
AND table1.id = table2.fieldD;

svn checkout -r <revision number>

This can be handy in case you use subversion in a production environment, where having the latest version is less important than having a version you know and can rely on.

There are numerous tutorials telling you that it would work, in version v1, v2 or v3 of the hardware - but they are all referring to the 32-bit version!

ndiswrapper uses the windows drivers, and since there are no proper 64bit XP drivers available, the device won’t work (yes, there are Vista 64 bit drivers, but ndiswrapper does not yet support Vista drivers). So be clever and give up, or install ubuntu 32 bit.

Tip: find out which hardware version your WG111 is:

$ lsusb

mine was “0846:4260″

I am using the older “-vhook” (video hook) functionality of ffmpeg. Please be carful with the location of the watermark.so library. Depending on your system, version of the library, etc. the location may differ. Look for it in either “/usr/lib/vhook/watermark.so” or “/usr/local/lib/vhook/watermark.so”!