I had the bright idea of watching the system calls during a paste with the strace utility, but all I got was 1444 lines of junk like this:

poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLOUT}])
writev(7, [{"7\5\4\0\312\341\0\2+Z\0\2\0\0\0\0007\4\4\0\313\341\0\2+Z\0\2\0\0\0\0008"..., 3920}, {"\1\1\1\1ZZZ_abbf\30\30\30\31\2\2\2\2\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 18864}, {""..., 0}], 3) = 22784
poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLIN|POLLOUT}])
read(7, "\34\0z\335\255\1\0\0L\1\0\0?\"?\4\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\34"..., 4096) = 64
writev(7, [{"<\5\2\0\332\341\0\2\226\10\t\0\3\0\0\0\331\341\0\2\0\0\0\0\325\341\0\2\0\0\0\0\0"..., 14864}, {"\0\0\0\0\n\n\v\17XYZ\202\213\214\215\316\232\233\235\344\230\230\232\340\224\225\227\333\221\222\223\326\215"..., 16704}, {""..., 0}], 3) = 31568
poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLIN}])
read(7, "\34\0\22\336\255\1\0\0\"\1\0\0B\"?\4\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096) = 32
poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLOUT}])
writev(7, [{"<\5\2\0\340\341\0\2\226\10\t\0\3\0\0\0\337\341\0\2\0\0\0\0\325\341\0\2\0\0\0\0\0"..., 16368}, {NULL, 0}, {""..., 0}], 3) = 16368
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLOUT}])
writev(7, [{"\226\10\t\0\3\341\0\2\364\341\0\2\0\0\0\0\370\341\0\2\0\0\0\0\0\0\0\0\0\0\4\0\4"..., 13656}, {"\1\1\1\1PQQTJKKN\20\20\20\20\1\1\1\1\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 2736}, {""..., 0}], 3) = 16392
clock_gettime(CLOCK_MONOTONIC, {71246, 415387992}) = 0
poll([{fd=3, events=POLLIN}, {fd=8, events=POLLIN}, {fd=9, events=POLLIN}, {fd=7, events=POLLIN}], 4, 0) = 0 (Timeout)
clock_gettime(CLOCK_MONOTONIC, {71246, 415497496}) = 0
poll([{fd=7, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=7, revents=POLLOUT}])
writev(7, [{"<\10\2\0#\342\0\2\226\10\t\0\3\0\0\0\"\342\0\2\0\0\0\0\36\342\0\2\0\0\0\0\0"..., 14764}, {NULL, 0}, {""..., 0}], 3) = 14764
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
clock_gettime(CLOCK_MONOTONIC, {71246, 416440091}) = 0
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=3, events=POLLIN}, {fd=8, events=POLLIN}, {fd=9, events=POLLIN}, {fd=7, events=POLLIN}], 4, 0) = 0 (Timeout)
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
clock_gettime(CLOCK_MONOTONIC, {71246, 418935003}) = 0
clock_gettime(CLOCK_MONOTONIC, {71246, 419022695}) = 0
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
clock_gettime(CLOCK_MONOTONIC, {71246, 419115259}) = 0
poll([{fd=3, events=POLLIN}, {fd=8, events=POLLIN}, {fd=9, events=POLLIN}, {fd=7, events=POLLIN}], 4, 4372) = 0 (Timeout)
read(7, 0x81557d0, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
clock_gettime(CLOCK_MONOTONIC, {71250, 795721149}) = 0
clock_gettime(CLOCK_MONOTONIC, {71250, 795764055}) = 0
clock_gettime(CLOCK_MONOTONIC, {71250, 795845848}) = 0

This is the one I have.

I find XA_CLIPBOARD in /usr/include/X11/Xmu/Atoms.h

Well, I guess that XA_CLIPBOARD is an extension of some kind then. I just thought it would be part of the core X11 library. I don't need it anyway, as XInternAtom(display , "CLIPBOARD" , true) finds it anyway.

I'm getting close to having a working clone of xclip that will have support for images, but it won't support images until I find out what format image editors are using to copy/paste images. Luckily my version of xclip will log what selection targets are available and are requested by other programs, so the more it gets run, the more information I will have.

Once my program is complete, I will bump the thread and post it.

Append :
Okay, here's my version of xclip, along with a version of ClipboardTest designed to use it. Linux users compile and test please.

ClipboardTest10.zip
You'll need to compile both programs separately this time and the command line should be something like :

gcc -O2 -Wall xcsi.c -LX11 -o xcsi.exe
gcc -O2 -Wall ClipboardTest.c Clipboard.c -o ClipboardTest.exe `allegro-config --libs`

To test, run ClipboardTest.exe. Copy text or an image in another program and then press the L key in ClipboardTest.exe. Do this for as many programs as you have that support copying text or images.

After that, try setting the clipboard string using S and getting the clipboard string using B.

When you're done, please post the contents of ClipboardLog.txt and xcsi_log.txt.

Thanks again for helping me test this.

Sorry, just erase the line in xcsi.c that #includes xcsi.h, I got rid of that file.

Append
Fixed the two things you mentioned and added in logging of the window name and target in selection requests.

ClipboardTest11.zip

Append2
m c, where did you go? Did you try the latest code?

It looks like Arthur's gone AWOL or on a walkabout or something...

@bamccaig || Trezker || Any other Linux users
Would one of you like to test things for me? Pretty please with sugar on top.

Okay, I made a diff using your altered xcsi.c and I updated mine to fix everything you pointed out. I had to change some of your changes though, so here's the new source code (ClipboardTest.c changed some as well) :

ClipboardTest12.zip

Now, if it quits, it should output why it quit if you run it from a console. I don't know if you need to specify a special flag with gcc to link it as a console application or not though. On MinGW, the flag is -Wl,--subsystem,console. So if it quits and you don't get any output, try that.

Also, thanks for compiling it. If I had Linux and gcc I wouldn't post code with so many errors in it.

You undid some of the things I did to get it to compile. I don't want to guess again.

pepsi@fractalcomet:/home/prog/ClipboardTest12 11:26 AM $ gcc -s -O2 -Wall xcsi.c -o xcsi.exe
xcsi.c: In function 'main':
xcsi.c:103: error: 'true' undeclared (first use in this function)
xcsi.c:103: error: (Each undeclared identifier is reported only once
xcsi.c:103: error: for each function it appears in.)
xcsi.c: In function 'QueryAvailableTargets':
xcsi.c:379: error: 'Event' undeclared (first use in this function)
xcsi.c:379: error: expected ';' before 'e'
xcsi.c:423: error: 'true' undeclared (first use in this function)
xcsi.c:446: error: 'e' undeclared (first use in this function)
xcsi.c: In function 'WaitForPaste':
xcsi.c:519: warning: unused variable 'return_property'
xcsi.c: In function 'RetrieveTarget':
xcsi.c:714: error: 'true' undeclared (first use in this function)
xcsi.c:730: error: expected ';' before 'if'
xcsi.c:689: warning: unused variable 'str'
xcsi.c:687: warning: unused variable 'xse'
pepsi@fractalcomet:/home/prog/ClipboardTest12 11:26 AM $ 

[EDIT]
I used yesterday's executable with this Clipboard.c and tried all the selections, got "0 Available Targets" in the console and this was in the logfile

RetrieveTarget warning : QueryAvailableTargets says this target is unavailable. Proceeding anyway.
RetrieveTarget failure : No selection owner.

OK, they all compiled this time.

Step by step test:
Run the clipboard executable.
Copy this line in a.cc reply box by dragging mouse.
Pressed B key and see the text appear correctly.
Pressed the S key and see the message about copying "CyberMan".
Pressed B key again to see "CyberMan" appear.
Pressed C key to see message "Could not set the clipboard bitmap to the CyberMan image"
Pressed V key to see message "There is no image in the clipboard to view".
Opened Cyberman_80x80.bmp with kolorpaint, select all and copy.
Pressed V key to see message "There is no image in the clipboard to view".
Pressed L key.

The resulting log file is in the paperclip.

[EDIT]

Virtual machines are pretty easy. You use the GUI to "create" a new machine,
you get a plethora of options as to the size of the virtual hard disk, how much
memory, to use the actual hardware DVD drive or to mount an ISO file, etc.
Then you either put the actual install disk for the OS into the drive or point
the virtual CD at the ISO, start the machine and install (in the window) just like
it was a real machine.
You may or may not have some trouble running hardware accelerated programs, but A4
and gcc should work fine. It will be slower than the real thing, obviously, but
not too bad.

I clicked on your avatar, right clicked to "View Image", then clicked on it again and selected "Copy Image" and ran the executable, only to have it say "There is no image in the clipboard to view" again. Log is in paperclip.

[EDIT]

You can't install hardware drivers to a virtual machine, it tries to emulate them with its own "drivers" but sometimes fails. It'll work with software rendering though, otherwise it couldn't even show a desktop.

That's what I used to do. It's not too bad to put up with depending in the BIOS and how often you switch. I did get tired of it though. With experience, I learned to just boot Windows from Grub, and make Grub my boot loader for all.

Depending on your distribution of choice (I generally use Fedora), the installer might even prompt you to add other partitions that it detects an operating system on (i.e., other Linux installs and even Windows). If not, it's relatively easy to manually add Windows to your grub.conf. After that, you modify the CMOS settings to always boot from your other hard drive (with Grub on it) which will provide you with a menu to choose from. You may need to alter the timeout though because IIRC Fedora 13's default timeout is 0, which basically means there's no time to actually choose an option. I changed it to like 3 (seconds) so there's enough time to interrupt it if you need to (the countdown stops if you interact with it), but it doesn't slow down booting too much. You can also pick the default boot entry, so if you want Windows to boot normally you can set that up.

I'm in Linux and don't use GUI tools for regular tasks. That's far too slow. I do it from the command line. I can (and do, from experience) view the contents of archives before extracting them, but it's always a pain when I find an archive that doesn't have a root directory; it means I need to create one, which slows me down. When I don't need to create one then I can generally just recall my last command, change a t or l to x and I'm done.

You promise it won't do anything malicious? (I don't feel like trying to make sense of all of the code )

** EDIT **

What am I missing?

[bamccaig@krypton ClipboardTest]$ gcc -Wall *.c `allegro-config --libs` -lX11
xcsi.c: In function ‘main’:
xcsi.c:95: warning: unused variable ‘j’
/tmp/ccetAgdd.o:(.bss+0x8): multiple definition of `logfile'
/tmp/ccoJCrKt.o:(.bss+0x0): first defined here
/tmp/ccetAgdd.o: In function `main':
xcsi.c:(.text+0x0): multiple definition of `main'
/tmp/cc1EN0iQ.o:ClipboardTest.c:(.text+0x13): first defined here
/tmp/ccetAgdd.o: In function `Log':
xcsi.c:(.text+0xb5b): multiple definition of `Log'
/tmp/ccoJCrKt.o:Clipboard.c:(.text+0x231): first defined here
/tmp/ccetAgdd.o: In function `CloseLog':
xcsi.c:(.text+0xba8): multiple definition of `CloseLog'
/tmp/ccoJCrKt.o:Clipboard.c:(.text+0x27e): first defined here
collect2: ld returned 1 exit status
[bamccaig@krypton ClipboardTest]$ 

This is from ClipboardTest13.zip.

I did

gcc -s -O2 -Wall xcsi.c -o xcsi
gcc -s -O2 ClipboardTest.c Clipboard.c -o clpbrd `allegro-config --libs`

clpbrd then exec()'s xcsi

My mistake was trying to compile everything into one... I wasn't really paying much attention (though this is what they invented README for).

As expected, it works to copy the string into the clipboard and retrieve it again. It fails to copy an image into or out of the clipboard.