Took me a while and some reading through the wacom sdk docs to figure out how to make it work in relative mode but in the end it only took these 3 lines of code added to EasyTab_Load()

LogContext.lcSysMode = 1; LogContext.lcSysSensX = 5000; LogContext.lcSysSensY = 5000;

However it would be nice if some information about this were added to the usage examples along with how to disable system mode if you don't want the tablet moving the cursor etc.

Cheers for the handy library btw =)

On handmade hero, @cmuratori is running into an issue where the tablet input in @serge-rgb's Milton stops working occasionally, and I think at one point he noticed it happen when his timer overlay thing gained focus.

I looked into the WinTab stuff and think calling WTEnable and WTOverlap when focus changes might fix it, but I don't have a tablet to test it.

Changing Wacom settings, like the wacom input area or screen area, when the EasyTab client app is open requires the client to handle it.

Changing display settings (like turning off one display) also requires the client to handle it, since the Wacom context needs to be rebuilt.

In my case, I am currently not handling it. My app needs a restart whenever those two particular cases happen. Not a huge deal, but the other applications I test seem to handle these cases without problems.

I was debating whether to open this issue or not, since this is one of those "Drop the Easy from EasyTab" issues.

The applications I mentioned that handle these cases seem to work when the wacom driver crashes, so I have to assume that they are using the Windows Stylus API. Also, the Wacom sample code is not helpful at all for cases like these.

So I'm thinking that one way forward should be to support the Stylus API and leave the Wacom layer as a fallback for when the Stylus API is not available or when EasyTab is compiled as C99.

This could add a lot of complexity to the code, so it might be a bad idea? I just wanted to write my thoughts somewhere that wasn't email, since this is an open source project. I'm curious to read your thoughts, but feel free to close the issue, since there is no code to show yet :)

I had a problem compiling EasyTab as a standalone lib (vs2015). It doesn't find calloc or strstr. On windows this is fixed by including #include <stdlib.h> and #include <string.h>.

IMO, the library should include what it needs, although in practice I am guessing that it will only ever be included into codebases that already have calloc and strstr included.

Related: Maybe instead of directly using calloc and free, there should be a EASYTAB_CALLOC and EASYTAB_FREE to give the client control over the allocation scheme.

1. Leave only Position and Pressure info in EasyTab. Which allows for it to be moved to the top where the public interface is declared.
2. Created EasyTabContext to hold the guts and live inside EASYTAB_IMPLEMENTATION.

I don't know if commit comments will be shown in the pull request. So:

• Removed MaxPressure from the public-facing struct. Pressure ranges in [0,1], so MaxPressure is currently only useful inside EasyTab_Handle() when doing the divide.
• Renamed struct member Context to WintabContext because EasyTabContext->Context was confusing

(Edit: Yup. Comments do show up in the Pull Request. Today I Learned :)

I would like to use this library however I do not feel comfortable using it until it has a license. Well I would like to believe the author means his claim of public domain it cannot hold up in every country and furthermore the author could turn around and sue for copyright. So I would like to propose adding a license of some sort. If you would like to keep it basically public domain I would recommend https://creativecommons.org/choose/zero/ as a license. It basically is public domain however covers more countries and ensures developer safety. At the end of the day it is your choice obviously but I would like to propose this.

Implemented a small fix to correct a memory leak when unloading EasyTab, the device handle returned by XOpenDevice was not being closed.

This does change the API, as unloading now requires a pointer to the Display* in order to close the device. I opted for this approach because it is more explicit and avoids retaining additional state. Alternatively, a copy to the original Display* passed in during load could be retained in the EasyTabInfo struct.

Cool project you started here! :)

Can you make a simple app that can be used to test tablets, like xev or xinput test on Linux, but for your library?

Thanks in advance!

EasyTab can't be currently linked as a library, because the declaration

static EasyTabInfo* EasyTab

creates two copies of EasyTab -- one in the library memory and one in the client memory.

Fix is to declare it as extern and define it once in the implementation code. (Pull request incoming)

When including easytab.h without defining EASYTAB_IMPLEMENTATION (defining it somewhere else for whatever reason), the global variable EasyTab is not available, so the client can't pull PosX, PosX, Pressure, etc...

(Sorry that I was the one who suggested moving the global in the first place)

Still... Should there be two structures? A lightweight struct that is exposed to the client and a heavyweight structure that contains all the guts?

I mean something like

typedef struct {
  int32_t PosX;
  int32_t PosY;
  float Pressure;
} EasyTab_;

static EasyTab_* EasyTab;

Right at the beginning of easytab.h, where the public interface is declared, and some other struct which contains the Win32 function pointers xor X11 stuff that can live beneath EASYTAB_IMPLEMENTATION.

• Stopped using bool and introduced EasyTabResult. It helped make it valid C99 and is a little step forward to having nice error reporting

• X11 had two problems. a) It was reporting in display coordinates, not window coordinates b) It was reporting in tablet resolution, not pixels (i.e. (x,y) in [0, EasyTab->RangeX]x[0, EasyTab->RangeY] which was different from the Windows behavior, where it works in pixels.

  Now, EasyTab_Handle is using X and Y window-relative coordinates reported by the XEvent. (This means that axis_data[0] and axis_data[1] is no longer used. Will open an issue about it)

  There was a bug (that was also present in Milton, so it's my bad) where it was not grabbing the correct window when asking for XDeviceMotionEvents (line 669). This changes the signature for EasyTab_Load on X11 but I'm pretty sure that it is the Right Way to do things.

[Edit: Clarity. Also, I had said something about EasyTab not working with multiple monitors, which wasn't true! EasyTab was working, but setting the whole monitor as the output area. ]

This implements support for screen area mapping, which means not setting the SysOrg and SysExt variables in LogContext.

Looking at lines 895 and 896 of the PR:

       Point.x = EasyTab->ScreenOriginX + Packet.pkX / EasyTab->ScreenAreaRatioX;
       Point.y = EasyTab->ScreenOriginY + Packet.pkY / EasyTab->ScreenAreaRatioY;

when before it was:

        Point.x = Packet.pkX;
        Point.y = Packet.pkY;

This should work as before when the feature is not being used.

• On a single monitor config, ScreenOrigin(X|Y) should be 0.
• When the feature is disabled, SysExt(X|Y) will be equal to CoordRange(X|Y), so ScreenAreaRatio(X|Y) will be 1.

I tested it on my machine enabling only one monitor and toggling the feature, but please test it if you choose to merge this :)

The feature works as expected with Milton compared with other programs that support it (Krita, Mischief, Gimp) but I have only tried it out with Milton. Maybe test it with Papaya first?

Finally, requesting code review :)