[PATCH] xserver-kdrive-common: fixed xTs_Calibrate script name.

[PATCH] xserver-kdrive-common: fixed xTs_Calibrate script name.

[Ihar Hrachyshka]

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 .../xserver-kdrive-common_0.1.bb                   |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb b/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
index 33c0a40..a6fbc6d 100644
--- a/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
+++ b/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
@@ -9,9 +9,9 @@ SRC_URI = "\
   file://Xserver \
   file://Xsession \
   \
-  file://30xTs_Calibrate \
   file://60xXDefaults \
   file://90xXWindowManager \
+  file://98xTs_Calibrate \
   "
 S = "${WORKDIR}"
 
@@ -22,9 +22,9 @@ etcFiles = "\
   Xsession \
   "
 sessionFiles = "\
-  30xTs_Calibrate \
   60xXDefaults \
   90xXWindowManager \
+  98xTs_Calibrate \
   "
 
 do_install() {
-- 
1.6.0.6

Re: [PATCH] xserver-kdrive-common: fixed xTs_Calibrate script name.

[Koen Kooi]

On 16-05-09 22:16, Ihar Hrachyshka wrote:
> Signed-off-by: Ihar Hrachyshka<ihar.hrachyshka@gmail.com>

Acked-by: Koen Kooi <koen@openembedded.org>


> ---
>   .../xserver-kdrive-common_0.1.bb                   |    4 ++--
>   1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb b/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
> index 33c0a40..a6fbc6d 100644
> --- a/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
> +++ b/recipes/xserver-kdrive-common/xserver-kdrive-common_0.1.bb
> @@ -9,9 +9,9 @@ SRC_URI = "\
>     file://Xserver \
>     file://Xsession \
>     \
> -  file://30xTs_Calibrate \
>     file://60xXDefaults \
>     file://90xXWindowManager \
> +  file://98xTs_Calibrate \
>     "
>   S = "${WORKDIR}"
>
> @@ -22,9 +22,9 @@ etcFiles = "\
>     Xsession \
>     "
>   sessionFiles = "\
> -  30xTs_Calibrate \
>     60xXDefaults \
>     90xXWindowManager \
> +  98xTs_Calibrate \
>     "
>
>   do_install() {

Re: [PATCH] xserver-kdrive-common: fixed xTs_Calibrate script name.

[Phil Blundell]

On Sat, 2009-05-16 at 23:16 +0300, Ihar Hrachyshka wrote:
>  sessionFiles = "\
> -  30xTs_Calibrate \
>    60xXDefaults \
>    90xXWindowManager \
> +  98xTs_Calibrate \
>    "

What's the reason for this?  Your email was a bit short on explanation
but it seems a little strange to be starting the window manager and
suchlike before calibration runs.

p.

Re: [PATCH] xserver-kdrive-common: fixed xTs_Calibrate script name.

[Florian Boor]

Hi,

Phil Blundell schrieb:
> What's the reason for this?  Your email was a bit short on explanation
> but it seems a little strange to be starting the window manager and
> suchlike before calibration runs.

I guess because I moved it in xserver-common as well. In fact there it still
runs before the window manager. Moving it after the window manager is a bug
(introduced by myself) and needs to be fixed. I'll do this now...

Greetings

Florian


-- 
The dream of yesterday                  Florian Boor
is the hope of today                    Tel: +49 271-771091-15
and the reality of tomorrow.            Fax: +49 271-771091-19
[Robert Hutchings Goddard, 1904]        florian.boor@kernelconcepts.de
                                        http://www.kernelconcepts.de/en

xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Looks like xtscal in x11-image fails.

$ DISPLAY=":0.0" xtscal
XCALIBRATE extension missing



When I look at /usr/lib/xorg/modules
I find
	libdri.la, but no libdri.so
	libglx.la, but no libglx.so

so they are not loaded.
Should they be there and loaded?


-------------------------------

Have gotten the touchscreen to work,
by

Section "ServerFlags"
	Option "AllowEmptyInput" "off"
EndSection
in xorg.conf

The touchscreen cursor travels in the x direction
when the pen moves in the y-direction,
but I think this is because the Linux config
specifies that the screen should be rotated 90',
which is does not do.
That is probably an fb driver problem,
which can be workaround by clearing the config.

------------------------------

But now the keyboard behaves in a strange manner
I get three chararcters per press.
I first through that the repeat rate was too
high, but the problem remained after
xset -r

I just realized that I get two characters
when I press the key, and the third character
when I release the key.

I think the driver reports key press, and something
else when I press, and key release when I release the button
and each event is interpreted as one character.

Any clues?

----------------------------------------------

BTW: You cannot set the repeat rate using
	Option "AutoRepeat" "500" "30"
since XFree86-Misc is no longer available.

"xset r rate 500" fails as well due to this.

Any alternative methods to turn off the repeat?










-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Phil Blundell]

On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
> Looks like xtscal in x11-image fails.
> 
> $ DISPLAY=":0.0" xtscal
> XCALIBRATE extension missing

That error is fairly self-explanatory: your xserver lacks the XCALIBRATE
extension.  Xtscal won't work if this is missing.

p.

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Phil Blundell wrote:
> On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
> > Looks like xtscal in x11-image fails.
> > 
> > $ DISPLAY=":0.0" xtscal
> > XCALIBRATE extension missing
> 
> That error is fairly self-explanatory: your xserver lacks the XCALIBRATE
> extension.  Xtscal won't work if this is missing.

Well, there may be a problem in recipes. There is xcalibrate and
libxcalibrate. Both provide the same library. Some recipes use
xcalibrate in DEPENDS, some use libxcalibrate. libxcalibrate does not
compile on images with xserver-kdrive.

I think that the fix will require complete removal of xcalibrate:
- xcalibrate_20060312.bb and xcalibrate_cvs.bb seems to be no more used.
- xcalibrate_git.bb needs to be moved to libxcalibrate_git.bb using its
version.
- libxcalibrate probably needs to be moved to model specific directories
(kdrive images need different version than xorg versions).
- Update preferred versions for particular image.
- and finally libxcalibrate should use standard git PV/PR allowing
future upgrade.

Alternative solution: Fix the latest libxcalibrate to work with bothe
Xorg and kdrive.


________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Stanislav Brabec skrev:
> Phil Blundell wrote:
>> On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
>>> Looks like xtscal in x11-image fails.
>>>
>>> $ DISPLAY=":0.0" xtscal
>>> XCALIBRATE extension missing
>> That error is fairly self-explanatory: your xserver lacks the XCALIBRATE
>> extension.  Xtscal won't work if this is missing.
> 
> Well, there may be a problem in recipes. There is xcalibrate and
> libxcalibrate. Both provide the same library. Some recipes use
> xcalibrate in DEPENDS, some use libxcalibrate. libxcalibrate does not
> compile on images with xserver-kdrive.
> 
> I think that the fix will require complete removal of xcalibrate:
> - xcalibrate_20060312.bb and xcalibrate_cvs.bb seems to be no more used.
> - xcalibrate_git.bb needs to be moved to libxcalibrate_git.bb using its
> version.
> - libxcalibrate probably needs to be moved to model specific directories
> (kdrive images need different version than xorg versions).
> - Update preferred versions for particular image.
> - and finally libxcalibrate should use standard git PV/PR allowing
> future upgrade.
> 
> Alternative solution: Fix the latest libxcalibrate to work with bothe
> Xorg and kdrive.


I have looked through the code.
x11-image uses Xorg and libxcalibrate.

There can be several reasons for this error message:

1) Xmalloc fails
2) There is no extension XCALIBRATE
3) There is an extension, but with different casing
   XCALIBRATE is used by libxcalibrate.
   if another name is used when it is added,
   then there is a problem when queried.
   The library is called libXcalibrate.so.

   How does it get its name?




xtscal contains:

  screen = DefaultScreen (dpy);

  if (XCalibrateQueryExtension (dpy, &event_base, &error_base))
    {
      int r;

      if (flag_debug)
	fprintf (stderr, "Using XCALIBRATE\n");

      r = XCalibrateSetRawMode (dpy, True);
      if (r)
	{
	  fprintf (stderr, "failed to set raw mode: error %d\n", r);
	  exit (1);
	}
    }
  else
    {
      perror ("XCALIBRATE extension missing");
      exit (1);
    }

The error message is printed, if XCalibrateQueryExtension returns false.
-----------------------------------------------------------------

I have found the XCalibrateQueryExtension in libxcalibrate/xcalibrate.c

Bool
XCalibrateQueryExtension (Display *dpy, int *event_basep, int *error_basep)
{
    XExtDisplayInfo *info = XCalibrateFindDisplay (dpy);

    if (XextHasExtension(info))
    {
	*event_basep = info->codes->first_event;
	*error_basep = info->codes->first_error;
	return True;
    }
    else
	return False;
}


XCalibrateQueryExtension returns false
	if XextHasExtension(info) is false
-----------------------------------------------------------------
From extutil.h:
#define XextHasExtension(i) ((i) && ((i)->codes))

=> ((info) && ((info)->codes)) is false

Problem occurs if
	info = XCalibrateFindDisplay (dpy)
results in
	info = NULL
	info->codes = NULL

-----------------------------------------------------------------
XCalibrateFindDisplay

XExtDisplayInfo *
XCalibrateFindDisplay (Display *dpy)
{
    XExtDisplayInfo *dpyinfo;
    XCalibrateInfo	    *xci;

    dpyinfo = XextFindDisplay (&XCalibrateExtensionInfo, dpy);
    if (!dpyinfo)
    {
	dpyinfo = XextAddDisplay (&XCalibrateExtensionInfo, dpy,
				  XCalibrateExtensionName,
				  &xcalibrate_extension_hooks,
				  XCalibrateNumberEvents, 0);
	xci = Xmalloc (sizeof (XCalibrateInfo));
	if (!xci)
	    return 0;
	xci->major_version = -1;
	dpyinfo->data = (char *) xci;
    }
    return dpyinfo;
}


if XextFindDisplay returns a non-NULL result, then this is returned.
	- not a problem

If it returns a NULL result, then
	XextAddDisplay is called.
	If the Xmalloc fails, we return 0	: Possible problem!
	if XextAddDisplay returns 0;		: Possible problem


-----------------------------------------------------------------
XextAddDisplay


If Xmalloc fails, we return 0.
If it does not fail, we set
	dpyinfo->codes = XInitExtension (dpy, ext_name);
	
if dpyinfo->codes == NULL we test
	if(hooks->close_display) ==
	if(&xcalibrate_extension_hooks->close_display) ==
	if(XCalibrateCloseDisplay) which I assume is true

	We then do
		codes = XAddExtension(dpy);
		if codes == non-NULL
			we return NULL 		-- Possible problem

	"XInitExtension()determines if the named extension exists. Then it
allocates storage for maintaining the information about the extension on
the connection, chains this onto the extension list for the connection,
and returns the information the stub implementor will need to access the
extension. If the extension does not exist, XInitExtension() returns 	NULL.
	
	If the extension name is not in the Host Portable Character Encoding
the result is implementation dependent. Case matters; the strings
thing, 	Thing, and thinG are all considered different names. "

	Do we have the right CASE for this name?
	If not we have a problem...

	char XCalibrateExtensionName[] = XCALIBRATE_NAME;

xcalibratewire.h
	#define XCALIBRATE_NAME "XCALIBRATE"

if dpyinfo->codes == non-NULL we return dpyinfo
	-- No problem

-------------------------------------------------------------
static /* const */ XExtensionHooks xcalibrate_extension_hooks = {
    NULL,				/* create_gc */
    NULL,				/* copy_gc */
    NULL,				/* flush_gc */
    NULL,				/* free_gc */
    NULL,				/* create_font */
    NULL,				/* free_font */
    XCalibrateCloseDisplay,		/* close_display */
    XCalibrateWireToEvent,		/* wire_to_event */
    XCalibrateEventToWire,		/* event_to_wire */
    NULL,				/* error */
    NULL,				/* error_string */
};
-------------------------------------------------------------


/*
 * XextAddDisplay - add a display to this extension
 */
XExtDisplayInfo *
XextAddDisplay(XExtensionInfo *extinfo, Display *dpy,
	       char *ext_name, XExtensionHooks *hooks, int nevents,
	       XPointer data)
{
    XExtDisplayInfo *dpyinfo;

    dpyinfo = (XExtDisplayInfo *) Xmalloc (sizeof (XExtDisplayInfo));
    if (!dpyinfo) return NULL;
    dpyinfo->display = dpy;
    dpyinfo->data = data;
    dpyinfo->codes = XInitExtension (dpy, ext_name);

    /*
     * if the server has the extension, then we can initialize the
     * appropriate function vectors
     */
    if (dpyinfo->codes) {
	int i, j;

	for (i = 0, j = dpyinfo->codes->first_event; i < nevents; i++, j++) {
	    XESetWireToEvent (dpy, j, hooks->wire_to_event);
	    XESetEventToWire (dpy, j, hooks->event_to_wire);
	}
	if (hooks->create_gc)
	  XESetCreateGC (dpy, dpyinfo->codes->extension, hooks->create_gc);
	if (hooks->copy_gc)
	  XESetCopyGC (dpy, dpyinfo->codes->extension, hooks->copy_gc);
	if (hooks->flush_gc)
	  XESetFlushGC (dpy, dpyinfo->codes->extension, hooks->flush_gc);
	if (hooks->free_gc)
	  XESetFreeGC (dpy, dpyinfo->codes->extension, hooks->free_gc);
	if (hooks->create_font)
	  XESetCreateFont (dpy, dpyinfo->codes->extension, hooks->create_font);
	if (hooks->free_font)
	  XESetFreeFont (dpy, dpyinfo->codes->extension, hooks->free_font);
	if (hooks->close_display)
	  XESetCloseDisplay (dpy, dpyinfo->codes->extension,
			     hooks->close_display);
	if (hooks->error)
	  XESetError (dpy, dpyinfo->codes->extension, hooks->error);
	if (hooks->error_string)
	  XESetErrorString (dpy, dpyinfo->codes->extension,
			    hooks->error_string);
    } else if (hooks->close_display) {
	/* The server doesn't have this extension.
	 * Use a private Xlib-internal extension to hang the close_display
	 * hook on so that the "cache" (extinfo->cur) is properly cleaned.
	 * (XBUG 7955)
	 */
	XExtCodes *codes = XAddExtension(dpy);
	if (!codes) {
	    XFree(dpyinfo);
	    return NULL;
	}
	XESetCloseDisplay (dpy, codes->extension, hooks->close_display);
    }

    /*
     * now, chain it onto the list
     */
    _XLockMutex(_Xglobal_lock);
    dpyinfo->next = extinfo->head;
    extinfo->head = dpyinfo;
    extinfo->cur = dpyinfo;
    extinfo->ndisplays++;
    _XUnlockMutex(_Xglobal_lock);
    return dpyinfo;
}

-------------------------------------------------------------
XAddExtension

> 
> 
> ________________________________________________________________________
> Stanislav Brabec
> http://www.penguin.cz/~utx/zaurus
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel


-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Ulf Samuelsson wrote:
> Stanislav Brabec skrev:
> > Well, there may be a problem in recipes. There is xcalibrate and
> > libxcalibrate. Both provide the same library. Some recipes use
> > xcalibrate in DEPENDS, some use libxcalibrate. libxcalibrate does not
> > compile on images with xserver-kdrive.
> > 
> I have looked through the code.
> x11-image uses Xorg and libxcalibrate.
> 
> There can be several reasons for this error message:
> 
> 1) Xmalloc fails
> 2) There is no extension XCALIBRATE
> 3) There is an extension, but with different casing
>    XCALIBRATE is used by libxcalibrate.
>    if another name is used when it is added,
>    then there is a problem when queried.
>    The library is called libXcalibrate.so.

4) Both xcalibrate and libxcalibrate were populated to staging and one
overwrites the other. .la files from one could "poison" the other.
To identify this possible problem, please look at your build logs or
check your tmp/work. If both xcalibrate and libxcalibrate were staged,
you may expect problem.

If only one of the was built, it is a different problem.


________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Stanislav Brabec skrev:
> Ulf Samuelsson wrote:
>> Stanislav Brabec skrev:
>>> Well, there may be a problem in recipes. There is xcalibrate and
>>> libxcalibrate. Both provide the same library. Some recipes use
>>> xcalibrate in DEPENDS, some use libxcalibrate. libxcalibrate does not
>>> compile on images with xserver-kdrive.
>>>
>> I have looked through the code.
>> x11-image uses Xorg and libxcalibrate.
>>
>> There can be several reasons for this error message:
>>
>> 1) Xmalloc fails
>> 2) There is no extension XCALIBRATE
>> 3) There is an extension, but with different casing
>>    XCALIBRATE is used by libxcalibrate.
>>    if another name is used when it is added,
>>    then there is a problem when queried.
>>    The library is called libXcalibrate.so.
> 
> 4) Both xcalibrate and libxcalibrate were populated to staging and one
> overwrites the other. .la files from one could "poison" the other.
> To identify this possible problem, please look at your build logs or
> check your tmp/work. If both xcalibrate and libxcalibrate were staged,
> you may expect problem.
> 
> If only one of the was built, it is a different problem.
> 
I did a patch to xtscal to print out all X11 Extensions
if the XInitExtension fails, and there is no XCALIBRATE
extension loaded.


The at91 confs have:
PREFERRED_PROVIDER_xserver = "xserver-kdrive"
This should be:
PREFERRED_PROVIDER_xserver = "xserver-kdrive-common"

Instead of kdrive, I am using the xorg server, and I saw
a comment that libXcalibrate does not work on xorg.

I will try with xerver-krive-common and see how it goes...


> 
> ________________________________________________________________________
> Stanislav Brabec
> http://www.penguin.cz/~utx/zaurus
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel


-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Phil Blundell]

On Sat, 2009-08-15 at 13:51 +0200, Ulf Samuelsson wrote:
> 3) There is an extension, but with different casing
>    XCALIBRATE is used by libxcalibrate.
>    if another name is used when it is added,
>    then there is a problem when queried.
>    The library is called libXcalibrate.so.
> 
>    How does it get its name?

There's no direct linkage between the name of the library and the name
of the extension: they're completely independent at a technical level,
although obviously the name of the library has been chosen to reflect
the extension that it relates to.  

The extension has always been named "XCALIBRATE" and it doesn't seem
very likely that its case has changed all of a sudden.  I don't expect
that this is the cause of your problem.

p.

Re: xtscal fails + other Xorg init problems.

[Phil Blundell]

On Sat, 2009-08-15 at 12:06 +0200, Stanislav Brabec wrote:
> Phil Blundell wrote:
> > On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
> > > Looks like xtscal in x11-image fails.
> > > 
> > > $ DISPLAY=":0.0" xtscal
> > > XCALIBRATE extension missing
> > 
> > That error is fairly self-explanatory: your xserver lacks the XCALIBRATE
> > extension.  Xtscal won't work if this is missing.
> 
> Well, there may be a problem in recipes. There is xcalibrate and
> libxcalibrate. Both provide the same library. Some recipes use
> xcalibrate in DEPENDS, some use libxcalibrate.

Although it might be true that there is an issue there, I don't think
this would be likely to cause the problem at hand.  If you get as far as
running xtscal then you must have at least some version of
libxcalibrate.so on hand, and as far as I know all recent-ish versions
of that library are wire-compatible so it shouldn't matter very much
which one you have.

It seems more likely that the xserver in question simply doesn't include
the extension.  Checking with xdpyinfo would quickly reveal whether or
not that's actually the case, anyway.

> libxcalibrate does not compile on images with xserver-kdrive.

I can't think offhand why libxcalibrate should care what xserver is in
your image.  What's the error you get?

p.

Re: xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Phil Blundell skrev:
> On Sat, 2009-08-15 at 12:06 +0200, Stanislav Brabec wrote:
>> Phil Blundell wrote:
>>> On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
>>>> Looks like xtscal in x11-image fails.
>>>>
>>>> $ DISPLAY=":0.0" xtscal
>>>> XCALIBRATE extension missing
>>> That error is fairly self-explanatory: your xserver lacks the XCALIBRATE
>>> extension.  Xtscal won't work if this is missing.
>> Well, there may be a problem in recipes. There is xcalibrate and
>> libxcalibrate. Both provide the same library. Some recipes use
>> xcalibrate in DEPENDS, some use libxcalibrate.
> 
> Although it might be true that there is an issue there, I don't think
> this would be likely to cause the problem at hand.  If you get as far as
> running xtscal then you must have at least some version of
> libxcalibrate.so on hand, and as far as I know all recent-ish versions
> of that library are wire-compatible so it shouldn't matter very much
> which one you have.
> 
> It seems more likely that the xserver in question simply doesn't include
> the extension.  Checking with xdpyinfo would quickly reveal whether or
> not that's actually the case, anyway.
> 
>> libxcalibrate does not compile on images with xserver-kdrive.
> 
> I can't think offhand why libxcalibrate should care what xserver is in
> your image.  What's the error you get?
> 
> p.
> 
This problem seems to be much simpler to solve than expected.
You just have to look at the right place

OE_EXTRA_CONF += "--enable_xcalibrate --enable-tslib"

in the Xserver's bb file!
Disabled by default.

Have not tested this yet, but if it walks like a duck ...



> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel


-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Phil Blundell wrote:
> On Sat, 2009-08-15 at 12:06 +0200, Stanislav Brabec wrote:
> 
> > libxcalibrate does not compile on images with xserver-kdrive.
> 
> I can't think offhand why libxcalibrate should care what xserver is in
> your image.  What's the error you get?

Well, I didn't it debug in deep yet, but I see:
| xcalibrate.c: In function 'XCalibrateScreenToCoord':
| xcalibrate.c:233: error: 'xXCalibrateScreenToCoordReq' undeclared (first use in this function)
| xcalibrate.c:233: error: (Each undeclared identifier is reported only once
| xcalibrate.c:233: error: for each function it appears in.)
| xcalibrate.c:233: error: 'req' undeclared (first use in this function)
| xcalibrate.c:234: error: 'xXCalibrateScreenToCoordReply' undeclared (first use in this function)
| xcalibrate.c:234: error: expected ';' before 'rep'
| xcalibrate.c:236: error: 'sz_xXCalibrateScreenToCoordReq' undeclared (first use in this function)
| xcalibrate.c:236: error: expected expression before ')' token
| xcalibrate.c:236: error: 'X_XCalibrateScreenToCoord' undeclared (first use in this function)
| xcalibrate.c:241: error: 'rep' undeclared (first use in this function)
| make: *** [xcalibrate.lo] Error 1
| FATAL: oe_runmake failed
NOTE: Task failed: /OE/build/tmp/work/armv5te-angstrom-linux-gnueabi/libxcalibrate-1_0.0+r0+gitr209d83af61ed38a002c8096377deac292b3e396c-r0/temp/log.do_compile.1481

Playing a bit with versions, recipes/xcalibrate/xcalibrate_git.bb is
built instead of recipes/xorg-lib/libxcalibrate_git.bb and the build
succeeds.

Notes:

Tested with patched libxcalibrate, that fixes its PV scheme.

zgrep -l xXCalibrateScreenToCoordReq ~/sources/*
git_anongit.freedesktop.org.git.xorg.lib.libXCalibrate_209d83af61ed38a002c8096377deac292b3e396c.tar.gz
git_anongit.freedesktop.org.git.xorg.proto.calibrateproto_1da6fd1e2c7a49648245c98481fabea8b9690a8c.tar.gz
xorg-server-1.6.1.tar.bz2


________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Stanislav Brabec wrote:
> Phil Blundell wrote:
> > On Sat, 2009-08-15 at 12:06 +0200, Stanislav Brabec wrote:
> > 
> > > libxcalibrate does not compile on images with xserver-kdrive.
> > 
> > I can't think offhand why libxcalibrate should care what xserver is
> in
> > your image.  What's the error you get?
> 

Well, the problem is located in the level between:

There are two recipes for the calibrate library: libxcalibrate and 
xcalibrate

There are two recipes that provide prototypes: calibrateproto 
and xcalibrabrateext

Only some of these combinations result in a working environment.

I'll test, whether latest library version works with xserver-
kdrive-1300.

-- 
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Stanislav Brabec wrote:

> There are two recipes for the calibrate library: libxcalibrate and 
> xcalibrate
> 
> There are two recipes that provide prototypes: calibrateproto 
> and xcalibrabrateext
> 
> Only some of these combinations result in a working environment.
> 
> I'll test, whether latest library version works with xserver-
> kdrive-1300.

Yes, xtscal works with xserver-kdrive-1300 and latest libxcalibrate and
calibrateproto.

It didn't work for kdrive based builds, because kdrive recipe wants
libxcalibrate, but xtscal wants xcalibrate. One dependency chain
overwrites the another with undefined result depending on the build
order.

That is why I would like to propose:
- remove all xcalibrate and xcalibrabrateext recipes
- find recipes and images that need them and migrate them to
libxcalibrate and calibrateproto.
- fix git versioning of libxcalibrate and calibrateproto, maybe
increment PE.


________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Koen Kooi]

On 18-08-09 12:26, Stanislav Brabec wrote:
> Stanislav Brabec wrote:
>
>> There are two recipes for the calibrate library: libxcalibrate and
>> xcalibrate
>>
>> There are two recipes that provide prototypes: calibrateproto
>> and xcalibrabrateext
>>
>> Only some of these combinations result in a working environment.
>>
>> I'll test, whether latest library version works with xserver-
>> kdrive-1300.
>
> Yes, xtscal works with xserver-kdrive-1300 and latest libxcalibrate and
> calibrateproto.
>
> It didn't work for kdrive based builds, because kdrive recipe wants
> libxcalibrate, but xtscal wants xcalibrate. One dependency chain
> overwrites the another with undefined result depending on the build
> order.
>
> That is why I would like to propose:
> - remove all xcalibrate and xcalibrabrateext recipes
> - find recipes and images that need them and migrate them to
> libxcalibrate and calibrateproto.
> - fix git versioning of libxcalibrate and calibrateproto, maybe
> increment PE.

Go ahead :)

Re: xtscal fails + other Xorg init problems.

[Phil Blundell]

On Tue, 2009-08-18 at 12:26 +0200, Stanislav Brabec wrote:
> That is why I would like to propose:
> - remove all xcalibrate and xcalibrabrateext recipes
> - find recipes and images that need them and migrate them to
> libxcalibrate and calibrateproto.
> - fix git versioning of libxcalibrate and calibrateproto, maybe
> increment PE.

Yes, that sounds reasonable enough.  I'm not quite sure how we ended up
with these multiple competing recipes for the same thing but clearly
there is no benefit to having them.  The only thing to watch out for
would be the backwards compatibility symlink for the pkgconfig data.

p.

Re: xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Stanislav Brabec skrev:
> Stanislav Brabec wrote:
> 
>> There are two recipes for the calibrate library: libxcalibrate and 
>> xcalibrate
>>
>> There are two recipes that provide prototypes: calibrateproto 
>> and xcalibrabrateext
>>
>> Only some of these combinations result in a working environment.
>>
>> I'll test, whether latest library version works with xserver-
>> kdrive-1300.
> 
> Yes, xtscal works with xserver-kdrive-1300 and latest libxcalibrate and
> calibrateproto.
> 
> It didn't work for kdrive based builds, because kdrive recipe wants
> libxcalibrate, but xtscal wants xcalibrate. One dependency chain
> overwrites the another with undefined result depending on the build
> order.

What is the difference between "xserver-kdrive-1300" and
"kdrive based builds" ?
Do you mean it does not work with later revisions of xserver-kdrive?



> 
> That is why I would like to propose:
> - remove all xcalibrate and xcalibrabrateext recipes
> - find recipes and images that need them and migrate them to
> libxcalibrate and calibrateproto.

Angstrom prefers another version of xserver-kdrive.

Should we try to focus on xorg-xserver-1.6.1 which seems to have kdrive
support?



> - fix git versioning of libxcalibrate and calibrateproto, maybe
> increment PE.
> 
> 
> ________________________________________________________________________
> Stanislav Brabec
> http://www.penguin.cz/~utx/zaurus
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel


-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Stanislav Brabec]

Ulf Samuelsson wrote:
> Stanislav Brabec skrev:
> > Stanislav Brabec wrote:

> > It didn't work for kdrive based builds, because kdrive recipe wants
> > libxcalibrate, but xtscal wants xcalibrate. One dependency chain
> > overwrites the another with undefined result depending on the build
> > order.
> 
> What is the difference between "xserver-kdrive-1300" and
> "kdrive based builds" ?

Just a package selection for particular target that causes different
order of overwriting of staged files.

> Do you mean it does not work with later revisions of xserver-kdrive?

xserver-kdrive-1300 is the oldest version in the tree. If the latest
libxcalibrate works with it, I guess that it will work with any newer
version.

> Should we try to focus on xorg-xserver-1.6.1 which seems to have kdrive
> support?

Are you sure that it is "kdrive support" and not just "kdrive in the
source tree", as it was in past (i. e. deprecated and without support)?

Well, anything newer than ancient xserver-kdrive-1300 working on
machines like spitz would be a win.

With a highly modularized Xorg, upgrade to full Xorg may be an option,
as it could support modern desktop extensions.

It also seems, that Xorg has much better performance in rotated fbdev
driver than xserver-kdrive-1300.

I was even able to run Xorg with OpenGL (Mesa emulated) on my Zaurus
(5-8FPS in glxgears).

Problems that block upgrage:

- broken keyboard driver (keyboard map loads only sometimes, some keys
or autorepeat doesn't load, kdrive > 1.3.0.0)
- support for different orientation of touchscreen and display (both
kdrive > 1.3.0.0 and xorg)
- working XRandR in fbdev driver (xorg, last time I tested it, just a
code for runtime rotation was missing, rotation itself was already
implemented)


________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: xtscal fails + other Xorg init problems.

[Ulf Samuelsson]

Stanislav Brabec skrev:
> Ulf Samuelsson wrote:
>> Stanislav Brabec skrev:
>>> Stanislav Brabec wrote:
> 
>>> It didn't work for kdrive based builds, because kdrive recipe wants
>>> libxcalibrate, but xtscal wants xcalibrate. One dependency chain
>>> overwrites the another with undefined result depending on the build
>>> order.
>> What is the difference between "xserver-kdrive-1300" and
>> "kdrive based builds" ?
> 
> Just a package selection for particular target that causes different
> order of overwriting of staged files.
> 
>> Do you mean it does not work with later revisions of xserver-kdrive?
> 
> xserver-kdrive-1300 is the oldest version in the tree. If the latest
> libxcalibrate works with it, I guess that it will work with any newer
> version.
> 
>> Should we try to focus on xorg-xserver-1.6.1 which seems to have kdrive
>> support?
> 
> Are you sure that it is "kdrive support" and not just "kdrive in the
> source tree", as it was in past (i. e. deprecated and without support)?
> 

No, I am on really deep water here.
This weekend was the first time I ever looked at X-Windows source.
Thankfully there are competent people here to answer questions.
Phil Blundell seems to have written most of the files I looked at!


> Well, anything newer than ancient xserver-kdrive-1300 working on
> machines like spitz would be a win.
> 
> With a highly modularized Xorg, upgrade to full Xorg may be an option,
> as it could support modern desktop extensions.
> 
> It also seems, that Xorg has much better performance in rotated fbdev
> driver than xserver-kdrive-1300.
> 
> I was even able to run Xorg with OpenGL (Mesa emulated) on my Zaurus
> (5-8FPS in glxgears).
> 
> Problems that block upgrage:
> 
> - broken keyboard driver (keyboard map loads only sometimes, some keys
> or autorepeat doesn't load, kdrive > 1.3.0.0)

I think I saw that Autorepeat requires an extension which has been removed

> - support for different orientation of touchscreen and display (both
> kdrive > 1.3.0.0 and xorg)
> - working XRandR in fbdev driver (xorg, last time I tested it, just a
> code for runtime rotation was missing, rotation itself was already
> implemented)
> 
> 
> ________________________________________________________________________
> Stanislav Brabec
> http://www.penguin.cz/~utx/zaurus
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel


-- 
Best Regards
Ulf Samuelsson

Re: xtscal fails + other Xorg init problems.

[Phil Blundell]

On Sat, 2009-08-15 at 11:15 +0200, Ulf Samuelsson wrote:
> When I look at /usr/lib/xorg/modules
> I find
> 	libdri.la, but no libdri.so
> 	libglx.la, but no libglx.so
> 
> so they are not loaded.
> Should they be there and loaded?

I think the .la files are definitely spurious: you wouldn't want them
installed on the target system under any circumstances I can think of.
As to whether you should have DRI and GLX in the first place, that
rather depends on your hardware capabilities.  Does a DRI driver exist
for your video card?

p.