tray | SharedPreferences replacement for Android with multiprocess | Runtime Evironment library

The drivers are bugged. Workarounds exists, but they are complex.

• Match driver name on Ricoh|HP and PCL6|PCL 6
• Filter any trayIds > 1000

Some drivers such as HP expose the printer trays properly in other areas, for example:

• HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Print\Printers\\PrinterDriverData
  • InputSlot
  • InputSlotDisplayNames

... however this isn't true for drivers such as Ricoh.

After examining a lot of drivers (HP, Ricoh, Xerox, Konica, etc) I've isolated the issue to the following:

• PCL6 drivers
• HP or Ricoh as the vendor

In cases with both vendors, the DC_BINS value is always > 1000, which is partially explained in the PCL6 MediaSource specification, quoting:

"... External input trays 1 through 248 are selected by substituting the values 8 through 255 for the enumerated values. Example, 8 = first external input tray, 9 = second external input tray, etc. ..."

Although there's nothing about the 1,000 specifically, and vendors such as Xerox use values over 7000 without the bug. That said, with the problematic vendors, what is observed is that when values are > 1,000 they tend to be actually valid MediaType values (NOT MediaSource values), but incremented by 1,000.

Oddly enough, this is very limited to HP and Ricoh and does not apply to other PCL drivers. For example:

• Konica uses trayID of 1000 = LCT, or "Large Capacity Tray", which is valid.
• Xerox offers a PCL6 driver, but commonly uses trayIDs higher than 1000, e.g. 7153 = Tray 1, 7154 = Tray 2.
• Ricoh is known to use trayID of 1025 in the PCL5 version of its driver, which is a valid tray value of 1025 = Auto Tray Select, but this doesn't seem to be the case for their PCL6 driver, which has the MediaType values mixed in.

So to "fix" this issue, I wrote a series of custom parsing to find out the driver vendor and the tray id.

To locate the trayId from Java:

Not a robust solution, it does work for the code currently posted but I do not assure this will work with the actual code you might test it on.

The regex expects a single string, hence, when testing this with your file, make sure you're using the -Raw switch.

See https://regex101.com/r/l0RLPw/2 for details.

This appears to have been fixed with release of version 2.14.6

Update: The following instructions work as advertised, but you state that you were aware of this approach, and that your only problem was that the styling of File Explorer's GUI in Windows 11 may falsely suggest that the window-style dropdown list (field Run in a shortcut file's Properties dialog) is disabled.

Here's how to run your script minimized on startup, via the user-specific "$env:APPDATA\Microsoft\Windows\Start Menu\Programs\Startup" folder.

• Move your script file, say foo.ps1 out of the startup folder,^[1] say to your desktop.

• In its place, create a shortcut file (.lnk) that calls your script file with its window minimized:

  • To do this interactively, via File Explorer's New > Shortcut shortcut-menu command:

    • Paste the following target command and submit:

You can create a new dataframe with the timestamp with fixed range of position. Then you merge them together and you will end up with NaN values on errors columns for given missing position. Then you fill the NaN to 1.

Sample code:

I recommend you to:

1. do camera calibration to make the lines in your image straight, and
2. align your camera well to your production line so you can simplify the image processing and make it more robust, and
3. use better illumination conditions if possible.

Given the images above, here is a brute force solution:

• you can first extract the region of interest (ROI) through: HSV segmentation followed by contour detection, then you can extract the rectangular shape of the tray using this SO answer.

EDIT 2022/2/28:
I have found a more elegant solution to this problem involving WTSQueryToken() and GetTokenInformation() thanks to this SO answer.

WTSQueryToken() returns a filtered user token, and without any extra steps, this token can be placed into a CreateProcessAsUserW() call to spawn my notification app.

I opt to call GetTokenInformation() to obtain the unfiltered (read: elevated) token for the user, which allows me to keep all my application's logs centralized in the Program Files directory. I am unclear if a process launched with an unfiltered token of a user without admin privileges will still be able to log to the Program Files directory, but this is non-essential to the process's ability to run.

SOLUTION:
The error message given for the thrown exception is Access is denied. This may or may not be the result of a bug related to the PROC_THREAD_ATTRIBUTE_PARENT_PROCESS flag. The workaround for this is simply to spawn the process again from the child process, using the exact same function call minus the problematic flag.

In my case, I pass a parameter that acts as a flag to tell the child process that it needs to relaunch itself.

EXPLANATION:
Thanks to the guidance of @IInspectable, I was able to resolve my issue.

Firstly, IInspectable informed me that the exception's data type was likely that of a winrt::hresult_error, which ended up being correct. From this, I was able to see that the exception message was Access is denied.

After updating my question to reflect this, IInspectable returned to point me towards the bottom of the comments of the Raymond Chen blog post that I linked in my question.

In short, the comment suggests that there is an underlying bug in the API that causes processes spawned in the way demonstrated by Chen to load with incorrect parameters, and offers a simple workaround by using Chen's function from within the child process to launch another child process. The comment also suggests that removing the PROC_THREAD_ATTRIBUTE_PARENT_PROCESS flag from the second function call is necessary, but this is not something I needed to do for the workaround to succeed.

For posterity, the comment from the blog post comments is as follows:

There are issues using the PROC_THREAD_ATTRIBUTE_PARENT_PROCESS flag… Environment variables for the new process are copied from the High-IL process not the Medium-IL process which cause issues with shell functions and some directory paths (%temp%, %userprofile% etc…) The token security descriptor for the process is also created without an ACE for the current user blocking access to most resources and the token security descriptor will also have a High-IL even though the process itself has a Medium-IL… This blocks the new process from accessing any system objects (events/pipes/sections/IPC etc…) while also blocking the process from opening its own process token with TOKEN_QUERY access. This seems to be a severe bug with the API but not sure if it’ll be fixed. As a workaround you can call CreateProcess a second time but from the new child process (without the PARENT_PROCESS flag) and it’ll be created with the correct token DAC security and environment variables (this is also why the above above sample doesn’t have issues with cmd.exe since it launches child processes).

You're using a pre-warmed image