I’ll try to reproduce on other platforms this afternoon

I can’t reproduce on MacOS (catalina) or linux (ARM RaspberryPi OS w/BC nightly) I don’t have Windows10 but I made a simple list-box% test https://gist.github.com/spdegabrielle/c32006baa6d285036bb5bae60600ccde

(it’t not really a test it just makes a big list-box for the purpose of determining if it is a problem with the control rather than the package manager GUI.)

I tested package manager on 7.6CS with Linunx and wasn’t able to reproduce.

The gist listbox works fine for me @spdegabrielle

Weird. I was assuming it was gui problem.

Does it only happen in ‘currently installed’

? What about the big list of packages?

(Next tab I can’t remember I’m in the kitchen )

Good question. I’ll try it. (Also try the “Currently Installed” again since the puter has had all night to “rest”)

Both “Currently Installed” and “Available from Catalog” stop part way through scrolling.

Can anyone run raco test -l tests/match/main and paste the result here? Racket’s CI says that all tests passed, but when I run it locally 6 tests fail.

Insteresting…

Can you also try to evaluate (match* (3 4)) and paste error message here (no need for context)?

> (match* (3 4)) ; match*: no matching clause for '(3 4) [,bt for context]

OK, something very weird is going on…

I’d be looking at the source (graphical package manager ?)

https://github.com/racket/racket/blob/master/pkgs/racket-test/tests/match/match-exn-tests.rkt#L34

This is the test that is failing for me

Yeah. I’ve taken the plunge.

And it looks like according to the result that you showed me, it should indeed fail

Wish me luck.

because it expects the text (3 4), but the error message has '(3 4)

And yet somehow the test passed on your computer…

Maybe the printer is different in the test suite?

Have you tried running that expression in the mzscheme language?

Ohh

That’s what’s going on

I changed the language

It won’t be bad. It’s not DrRacket.

Haha

Resizing the window makes it not get stuck

Woah, that’s impressive debugging skills right there @samth! :clap:

On some window sizes it doesn’t get stuck scrolling down but once you hit bottom and scroll back up it gets stuck at the same spot scrolling up

Well - he has an unfair advantage. ;-) He wrote match .

Honestly, one of my weird talents is figuring out errors without doing any work (for other people, it never works on my bugs)

For this one, I just noticed that the problem was quote-printing, clicked the link, saw that it used mzscheme, and then hypothesized

Actually, I didn’t change mzscheme on that file itself. I changed scheme/base to racket/base in the entry file (https://github.com/racket/racket/blob/master/pkgs/racket-test/tests/match/main.rkt)

Didn’t mzscheme use an older match?

mzlib/match denotes a different pattern syntax than racket/match

That’s a very useful talent!

Btw, @laurent.orseau, I just make (recompile Racket from source) and it shows this error to me:

raco setup: 1 rendering: <pkgs>/quickscript/scribblings/quickscript.scrbl open-input-file: cannot open input file path: /Users/sorawee/git/racket/racket/share/pkgs/quickscript/img/warning.png system error: No such file or directory; errno=2 context...:

I think you might need define-runtime-path?

yes, looks like it :disappointed:

thanks for the quick report!

Initially I had used a utf8 char, but it broke the nightly build… So I just removed it altogether now, it caused too much pain already.

(pushed)

Annoying that this doesn’t raise an error locally

it will raise an error if you run it in a different directory

OK. I was able to get the dimensions of just the list-box% control when it was wonky and when I use those values for the min-width and min-height in your simpler gist code I’m able to reproduce the problem.

:tada:

Seems to behave weird with: [min-width 780] [min-height 485]

The width most likely doesn’t matter.

Only on win10?

Yeah I checked with my Linux box with those sizes and it works fine. Now peeking at the wx list box code.

Through some strategic display usage I’ve found it only happens when there are a multiple of the scroll step visible.

So default is 3 so it happens with 12, 15, 18, 21, etc items

Alirghty

https://github.com/racket/gui/blob/master/gui-lib/mred/private/wx/win32/list-box.rkt#L414

I think that’s why it doesn’t work. If the item we are trying to set as the top item is the same as the number of visible items it doesn’t actually move the list.

Not sure what the reasoning behind that is

If the scroll preference was set to 1 it would happen on every list box

So the code to stop scrolling when the scroll box is small is also triggering in for multiples?

Sorry not at a computer so can’t look at the code.

I’m using call/input-url so the port is closed automatically for me and Im using linux

Can anyone help me figure out what’s wrong with this syntax class? Trying to use the enum-id.predicate attribute raises enum-descriptor-predicate: unbound identifier; also, no #%app syntax transformer is bound. #lang racket/base (require racket/contract/base) (provide enum-id (contract-out [enum-binding (-> enum-type? identifier? enum-binding?)])) (require (for-label racket/base rebellion/type/enum/descriptor) rebellion/type/enum/base syntax/parse) ;@------------------------------------------------------------------------------ (struct enum-binding (type descriptor)) (define-syntax-class enum-id #:attributes (type descriptor predicate) (pattern binding #:declare binding (static enum-binding? "a static enum binding") #:attr type (enum-binding-type (attribute binding.value)) #:with descriptor (enum-binding-descriptor (attribute binding.value)) #:with predicate #'(enum-descriptor-predicate descriptor)))

(I’m on Racket BC 7.7)

Guessing from my phone: the no app% error often happens when some require is missing in the syntax phase, so missing a for-syntax of something?

tried that, didn’t work :(

logically I think I have everything needed, because #' is being used at phase 0 and enum-descriptor-predicate is being used at phase –1, and I have (for-label racket/base rebellion/type/enum/descriptor).

dammit I just realized the problem: I used for-label but I meant for-template

Update: fixed the bug. The for-label is supposed to be for-template.

Well you have two directions: try adding things until it works, or try removing things until it works ;)

That was neither haha (changing things)