Re: wintab: Incorrectly classifying cursor as stylus instead of eraser
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
Did not attach the patch in my last email :(
For Wacom Pen & Touch Intuos tablet eraser cursor is incorrectly detected as stylus. The tablet returns its eraser cursor name as: Wacom Intuos PT S Pen eraser and type as: ERASER
Current code classifies this as STYLUS, because name of the cursor contains the word 'Pen'. It is very unlikely that a cursor with type ERASER is STYLUS.
There is already a check for eraser type, you should be able to use that instead. -- Alexandre Julliard julliard(a)winehq.org
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER; I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better? On 2014.03.20. 22:11, Alexandre Julliard wrote:
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
Did not attach the patch in my last email :(
For Wacom Pen & Touch Intuos tablet eraser cursor is incorrectly detected as stylus. The tablet returns its eraser cursor name as: Wacom Intuos PT S Pen eraser and type as: ERASER
Current code classifies this as STYLUS, because name of the cursor contains the word 'Pen'. It is very unlikely that a cursor with type ERASER is STYLUS. There is already a check for eraser type, you should be able to use that instead.
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER;
I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better?
Maybe a better way would be to restructure the checks, and give priority to the type over the name or something like that. The thing is that we don't want to have an is_stylus function that also checks for erasers, and then an is_eraser function with slightly different checks. -- Alexandre Julliard julliard(a)winehq.org
To be able to prioritize we have to change current functions into smth like this: get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; } and rewriting is_stylus and is_eraser into more simple functions like is_stylus(cursor) { return get_cursor_type(cursor) == CSR_TYPE_PEN; } On 2014.03.21. 10:54, Alexandre Julliard wrote:
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER;
I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better? Maybe a better way would be to restructure the checks, and give priority to the type over the name or something like that. The thing is that we don't want to have an is_stylus function that also checks for erasers, and then an is_eraser function with slightly different checks.
On 03/21/2014 10:04 AM, Eriks Dobelis wrote:
To be able to prioritize we have to change current functions into smth like this:
get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; }
and rewriting is_stylus and is_eraser into more simple functions like is_stylus(cursor) { return get_cursor_type(cursor) == CSR_TYPE_PEN; } In that case you could actually get rid of the is_* functions; "get_cursor_type(cursor) == CSR_TYPE_PEN" is as readable.
bye michael
On 2014.03.21. 10:54, Alexandre Julliard wrote:
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER;
I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better? Maybe a better way would be to restructure the checks, and give priority to the type over the name or something like that. The thing is that we don't want to have an is_stylus function that also checks for erasers, and then an is_eraser function with slightly different checks.
-- Michael Stefaniuc Tel.: +49-711-96437-199 Supervisor, APAC/EMEA IT Network Fax.: +49-711-96437-111 -------------------------------------------------------------------- Reg. Adresse: Red Hat GmbH, Werner-von-Siemens-Ring 14, 85630 Grasbrunn Handelsregister: Amtsgericht Muenchen HRB 153243 Geschäftsführer: Mark Hegarty, Charlie Peters, Michael Cunningham, Charles Cachera
Hello,
To be able to prioritize we have to change current functions into smth like this:
get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; }
I worked on this code a little bit back in the day, I remember the detection stuff getting weird due to some odd ball drivers (non-wacom stuff doing strange things). Bug http://bugs.winehq.org/show_bug.cgi?id=11699 and specifically the log http://bugs.winehq.org/attachment.cgi?id=13358 of an Acecad table driver's settings. Just thought it might be handy to consider if you were to take a shot at rewriting the detection code. Regards, --John On Fri, Mar 21, 2014 at 4:09 AM, Michael Stefaniuc <mstefani(a)redhat.com>wrote:
On 03/21/2014 10:04 AM, Eriks Dobelis wrote:
To be able to prioritize we have to change current functions into smth like this:
get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; }
and rewriting is_stylus and is_eraser into more simple functions like is_stylus(cursor) { return get_cursor_type(cursor) == CSR_TYPE_PEN; } In that case you could actually get rid of the is_* functions; "get_cursor_type(cursor) == CSR_TYPE_PEN" is as readable.
bye michael
On 2014.03.21. 10:54, Alexandre Julliard wrote:
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER;
I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better? Maybe a better way would be to restructure the checks, and give priority to the type over the name or something like that. The thing is that we don't want to have an is_stylus function that also checks for erasers, and then an is_eraser function with slightly different checks.
-- Michael Stefaniuc Tel.: +49-711-96437-199 Supervisor, APAC/EMEA IT Network Fax.: +49-711-96437-111 -------------------------------------------------------------------- Reg. Adresse: Red Hat GmbH, Werner-von-Siemens-Ring 14, 85630 Grasbrunn Handelsregister: Amtsgericht Muenchen HRB 153243 Geschäftsführer: Mark Hegarty, Charlie Peters, Michael Cunningham, Charles Cachera
Found another oddball log: http://www.winehq.org/pipermail/wine-devel/2013-May/099905.html I don't have any suggestions on what the right solution is... just wanted to make sure you had all the logs. Regards, --John On Sat, Mar 22, 2014 at 12:28 PM, John Klehm <xixsimplicityxix(a)gmail.com>wrote:
Hello,
To be able to prioritize we have to change current functions into smth like this:
get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; }
I worked on this code a little bit back in the day, I remember the detection stuff getting weird due to some odd ball drivers (non-wacom stuff doing strange things).
Bug http://bugs.winehq.org/show_bug.cgi?id=11699 and specifically the log http://bugs.winehq.org/attachment.cgi?id=13358 of an Acecad table driver's settings.
Just thought it might be handy to consider if you were to take a shot at rewriting the detection code.
Regards, --John
On Fri, Mar 21, 2014 at 4:09 AM, Michael Stefaniuc <mstefani(a)redhat.com>wrote:
On 03/21/2014 10:04 AM, Eriks Dobelis wrote:
To be able to prioritize we have to change current functions into smth like this:
get_cursor_type(cursor) { //new function can we determine type based on cursor.device? Yes -> return CSR_TYPE_PEN; can we determine type based on cursor.name? Yes -> return CSR_TYPE_ERASER; return CSR_TYPE_OTHER; }
and rewriting is_stylus and is_eraser into more simple functions like is_stylus(cursor) { return get_cursor_type(cursor) == CSR_TYPE_PEN; } In that case you could actually get rid of the is_* functions; "get_cursor_type(cursor) == CSR_TYPE_PEN" is as readable.
bye michael
On 2014.03.21. 10:54, Alexandre Julliard wrote:
Eriks Dobelis <eriks.dobelis(a)biti.lv> writes:
The alternative - just to switch order of checks in X11DRV_LoadTabletInfo. There is a code fragment: /* prefer finding TYPE_PEN(most capable) */ if (is_stylus(target->name, device_type)) cursor.TYPE = CSR_TYPE_PEN; else if (is_eraser(target->name, device_type)) cursor.TYPE = CSR_TYPE_ERASER; else cursor.TYPE = CSR_TYPE_OTHER;
I would need to switch order and check for ERASER first. Problems would arise if for some device name would be "Pen & Eraser", and type would be "STYLUS" , and then it would be detected as eraser. In the proposed patch this problem would not arise as only device is checked for ERASER. For my specific case both approaches work. Which one you consider as better? Maybe a better way would be to restructure the checks, and give
priority
to the type over the name or something like that. The thing is that we don't want to have an is_stylus function that also checks for erasers, and then an is_eraser function with slightly different checks.
-- Michael Stefaniuc Tel.: +49-711-96437-199 Supervisor, APAC/EMEA IT Network Fax.: +49-711-96437-111 -------------------------------------------------------------------- Reg. Adresse: Red Hat GmbH, Werner-von-Siemens-Ring 14, 85630 Grasbrunn Handelsregister: Amtsgericht Muenchen HRB 153243 Geschäftsführer: Mark Hegarty, Charlie Peters, Michael Cunningham, Charles Cachera
participants (4)
-
Alexandre Julliard -
Eriks Dobelis -
John Klehm -
Michael Stefaniuc