cocoa,canvas: fix bezel-cell frame for FocusView #326
Conversation
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Resyntax analyzed 1 file in this pull request and found no issues.
|
Thanks!! |
|
Did this get into Racket v8.13? |
|
It probably doesn't matter anyway, right? The dmg package is unaffected. It affects only build from source, but in that case you can just pull a fixed version instead. |
|
In our class, we have trouble running |
|
A problem with
|
On the student's machine (macOS Sonoma, x86_64, Racket CS v8.12), the big-bang window is entirely blank and sometimes it gets painted with the content of the Interaction Window. This happens consistently at least on that particular machine. I have not checked with others to see if this happens for other people.
Sort of, because nothing ever shows up in the
I'll try to find the student. |
|
We're considering this as a fix that might need to go into the 8.13 release (which is halfway out the door) or if it can wait for 8.14 (in 3 months, give or take). Do folks have any opinion on this point? @jbclements |
|
Actually, I realize that this is not urgent for v8.13. Assuming that v8.14 is going to be released in August, the fix can make it into Fall quarter/semester. |
|
I'm seeing somewhat conflicting information about which processors & operating systems are affected. It appears that it's only macOS 14, Sonoma, but is it only x86_64 or does it affect ARM as well? Also, I'm hoping someone can confirm that this bug was present in 8.12 as well? The rather visible effect in the preference pane is distressing, to be sure. |
|
On AArch64, I believe the bug affects only users who build Racket from source. It doesn't affect users who run a release or snapshot distribution. (It's not so much about a release/snapshot being built on an old version of the OS as being built for older versions of the OS by selecting the 11.1 SDK. I have seen this effect one other time: racket/racket@eccc500.) I am unclear on the status for x86_64. I have seen older release distributions of the x86_64 executable run a big-bang-like program badly on an AArch64 machine via Rosetta, and I had chalked that up to a Rosetta issue, but maybe it was the same bug. In any case, that's not the same as running on an x86_64 Sonoma installation. |
|
If we're going to test this on a variety of old macs we have lying around, is there a particular program to run that we can compare before/after the fix with? |
|
Here is the code that's not working on the student's machine. With the code in the ZIP file, it should be showing a static, non-moving game. The same code with the snapshot build works. |
|
@shhyou just to be clear, did the student build their Racket from source? Or ran Racket via Rosetta? I don't think "the snapshot build works" tells anything, because the snapshot build has always been working fine (even before this PR). |
|
No, the student uses v8.12 dmg and is on an x86_64 machine. |
|
@shhyou I'm not sure how to download the code that you posted. I see the text "Exercise 5 Snake.zip (see edit history)", but I'm not sure how to interpret that, or to download the code. I'm also wondering whether it isn't sufficient to run the code that @Bogdanp provided in the bug report, to wit: #lang racket/gui
(define frame
(new frame%
[label "Test"]))
(define msg
(new message%
[label "test"]
[parent frame]))
(define input
(new text-field%
[label "test"]
[parent frame]))
(send frame show #t)
|
|
Also, there seems to be some conflict between Matthew's suggestion that this does not affect snapshot or release builds (because they target the 11.1 SDK), and @shhyou 's evidence of a student that is apparently having this problem using the 8.12 release dmg. (EDIT: Okay, I see that Matthew leaves the door open for a different possible behavior on x86_64 systems.) FWIW, I just tested the release bundle on an x86_64 mac running macOS Monterey (12.7.4), and unsurprisingly, I don't see the problem. |
|
Okay... I just tried to upgrade my x86_64 mac to Sonoma, and Sonoma simply isn't available for my (2015) x86_64 machine. Ah well. |
I've sent it offline, just in case. |
|
I'd like to help test but I'm not sure if what I have is useful. It is a late 2013 imac running high sierra (10.13). It doesn't have the problem with the snake game in 8.12, on either of the 32 or 64 bit builds. I somehow have the (possibly incorrect?) impression that it might be a useful machine to test a fixed Racket on for the release. I'm happy to do so! @jbclements there is a little popdown menu near the word "edited" in @shhyou 's message with the snake game; if you go to the first version of the message, you'll be able to download the game. At least, that worked for me. |
I updated an x86_64 machine I have to sonoma and used the x8.12 dmg with the snake game code posted here earlier and I didn't see any problems. (I'm surprised by this ...) |
Fixes #325. The
rargument represents the portion of the view that is "dirty", and it seems like it may be larger than the view's frame[1]. It looks like this code was already intending to use the view's frame to draw the bezel, but accidentally used the dirty rect instead.[1]: At least on recent versions of macOS, which may explain why this isn't a problem when using versions of Racket built on older macOS versions.