public inbox for cygwin@cygwin.com
 help / color / mirror / Atom feed
From: Jon Turney <jon.turney@dronecode.org.uk>
To: Ken Whitesell <KenWhitesell@comcast.net>,
	The Cygwin Mailing List <cygwin@cygwin.com>
Subject: Re: Cygwin/X with Win10 display scaling corrupting font display of typed characters - Issue identified - "Solution" found
Date: Sat, 5 Feb 2022 14:25:48 +0000	[thread overview]
Message-ID: <b5d8769c-cf7c-e654-83c0-b7f375e5f64b@dronecode.org.uk> (raw)
In-Reply-To: <2e8f1d16-92eb-b66e-c49e-7e38e5f07110@comcast.net>

On 27/01/2022 03:12, Ken Whitesell wrote:
> On 1/24/2022 10:02 AM, Jon Turney wrote:
>> On 20/01/2022 01:01, Ken Whitesell wrote:
>>> On 1/19/2022 2:28 PM, Jon Turney wrote:
>>>> On 19/01/2022 00:02, Ken Whitesell wrote:
>>>>> On 1/17/2022 1:29 PM, Ken Whitesell wrote:
>>>>>>
>>>>>> Is there a known solution for this? (Or is it known that there is 
>>>>>> no solution?)
>>>>
>>>> Thanks for reporting this.
>>>>
>>>>>> Any guidance, pointers, suggestions of avenues for further 
>>>>>> research, or other information, will all be greatly appreciated.
>>>>>>
>>>>> After more research and experimentation, it appears to be related 
>>>>> to one of xorg-server, xorg-server-common, or xorg-server-xorg.
>>>>>
>>>>> Installing the older version 1.20.12-1 of these packages allows the 
>>>>> windows to be moved between monitors without any issues. Upgrading 
>>>>> to the current version 21.1.3-1 creates the problems. I'm able to 
>>>>> replicate this behavior on two different laptops with two different 
>>>>> external monitors.
>>>>
>>>> It seems likely that this is an unintended effect of changes in 
>>>> xorg-server 21.1.0-1, trying to fix problems in this area (See [1])
>>>
>>> Thanks for the references. I've read all the messages in the thread - 
>>> I was particularly intrigued by this comment:
>>>
>>> wrt the font scaling issue, looking at the source, it seems that we
>>> don't re-consider the display dpi after a WM_DISPLAYCHANGE message, but
>>> keep on using the value determined at startup.  This is probably a bug.
>>>
>>> I'm curious enough to want to take a look at the code, but I've got 
>>> no belief that I'm going to be able to find an answer. (I'm *not* a 
>>> C++ programmer. I can read it and write a little of it, but that's 
>>> about it.) I was going to start by comparing the last known-working 
>>> version to the first known-non-working version, but given that it's a 
>>> major release change, that's not likely going to be a useful 
>>> approach. (I'm way out of my league here. It's probably going to take 
>>> me a long time just to get to the point where I can even begin to 
>>> explore this.)
>>
>> The relevant change, which tries to fix the issue identified in that 
>> comment, and probably introduces this issue is:
>>
>> https://gitlab.freedesktop.org/jturney/xserver/-/commit/b19b6266d33f2b911dc1826ad5c03da135a39957 
>>
>>
>> [...]
[...]
> 
> But, the real reason for my reply at this point is to report that I have 
> found a solution for _my_ issue.
> 
> I make no guarantees about any problems anyone else may be facing, nor 
> can I make any statement about whether or not this causes other problems.
> 
> Obligatory disclaimer: I don't really know what I'm doing here. I'm not 
> a Windows developer, and I know just enough about cygwin to muddle my 
> way through doing what I want to do with the help of the mailing lists 
> and other resources. I'm not in a position to _explain_ anything. This 
> works for me, but that's as far as I can go.
> 
> First, the bottom line:
> 
> XWin.exe.manifest, line 21
> 
> change:
> <dpiAwareness>PerMonitorV2,PerMonitor</dpiAwareness>
> to
> <dpiAwareness>PerMonitor</dpiAwareness>
> 
> Some details:
> 
> I managed to get to a point where I could build the packages from source 
> and install them. I looked at the commit you referred me to, and started 
> reverting changes, one-by-one - at least in so far as the change 
> appeared to make sense to me.
> 
> Anyway, I got to this change, and sure enough, it worked. Removing the 
> "PerMonitorV2" solved the issue. Also, I confirmed that it's the 
> "PerMonitorV2" that is causing the issue and not having both of them by 
> running another test with just the "PerMonitorV2" - and that still shows 
> the problem.

Thanks for taking the time to narrow this down.  It's been very helpful.

Working through the documented effects of that [1], I was able to work 
out that this mis-rendering is due to the non-client area scaling.

[1] See under DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2 in
https://docs.microsoft.com/en-us/windows/win32/hidpi/dpi-awareness-context

If you turn off 'Hide Root Window' from the tray menu, you can kind of 
see what's going wrong: normally the client area of the X window is 
exactly aligned with that in the root window, but when the non-client 
area is rescaled by a DPI change (the title bar changes size 
significantly), it's misaligned so that part of the X window is outside 
the client area (and thus clipped in updates).

I think this is due to the AdjustWindowsRectEx() function not being DPI 
aware (I guess it's always computing the non-client window rect based on 
the processes initial DPI)

Unfortunately, from an initial look, rewriting things to use 
AdjustWindowRectExForDpi() isn't trivial (since we need to make 'DPI of 
the monitor this Window is going to end up on' available to it)

So for the moment, I think I'll apply your reversion (although this 
probably comes at the cost of not scaling the window frame, the traymenu 
and About... dialog)


  reply	other threads:[~2022-02-05 14:26 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-01-17 18:29 Cygwin/X with Win10 display scaling corrupting font display of typed characters Ken Whitesell
2022-01-19  0:02 ` Cygwin/X with Win10 display scaling corrupting font display of typed characters - Issue identified Ken Whitesell
2022-01-19 19:28   ` Jon Turney
2022-01-20  1:01     ` Ken Whitesell
2022-01-21 18:26       ` Cygwin/X with Win[10]-^ display scaling corrupting font display of typed characters - Issue [identified]-???? L A Walsh
2022-01-23 20:25         ` Cygwin/X display scaling corrupting font display of typed characters; L A Walsh
2022-01-24 15:02       ` Cygwin/X with Win10 display scaling corrupting font display of typed characters - Issue identified Jon Turney
2022-01-27  3:12         ` Cygwin/X with Win10 display scaling corrupting font display of typed characters - Issue identified - "Solution" found Ken Whitesell
2022-02-05 14:25           ` Jon Turney [this message]
2022-08-14 11:08             ` Jon Turney

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b5d8769c-cf7c-e654-83c0-b7f375e5f64b@dronecode.org.uk \
    --to=jon.turney@dronecode.org.uk \
    --cc=KenWhitesell@comcast.net \
    --cc=cygwin@cygwin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).