# Windows, GTK4, printRunPageSetupDialog problem

**URL:** <https://discourse.haskell.org/t/windows-gtk4-printrunpagesetupdialog-problem/10869>\
**Category:** Learn\
**Created:** [December 1, 2024, 6:10pm UTC](https://discourse.haskell.org/t/windows-gtk4-printrunpagesetupdialog-problem/10869 "2024-12-01T18:10:05Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![mpilgrem](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.haskell.org/mpilgrem/32/2529_2.png) [@mpilgrem](https://discourse.haskell.org/u/mpilgrem)\
**Post date:** [December 1, 2024, 6:10pm UTC](https://discourse.haskell.org/t/windows-gtk4-printrunpagesetupdialog-problem/10869/1 "2024-12-01T18:10:05Z")

</div>

My Haskell/GTK4 experiments have reached ‘printing’. I have a run time problem which may be a GTK4 bug, but I am seeking to rule out the Haskell dimension first. I have a simple application which compiles with GHC 9.6.6:

```haskell
{-# LANGUAGE ImplicitParams #-}
{-# LANGUAGE OverloadedLabels #-}
{-# LANGUAGE OverloadedRecordDot #-}
{-# LANGUAGE OverloadedStrings #-}

module Main
  ( main
  ) where

import Control.Monad ( void )
import Data.GI.Base ( AttrOp (..), new, on )
import qualified GI.Gtk as Gtk

main :: IO ()
main = do
  app <- new Gtk.Application
    [ #applicationId := "com.pilgrem.gtk-print"
    , On #activate (activate ?self)
    ]
  void $ app.run Nothing

activate :: Gtk.Application -> IO ()
activate app = do
  window <- new Gtk.ApplicationWindow
    [ #application := app
    , #title := "GTK4 printer example"
    , #resizable := False
    ]

  pageSetupButton <- new Gtk.Button
    [#label := "Page setup"]

  let onClickPageSetup :: IO ()
      onClickPageSetup = do
        printSettings <- Gtk.printSettingsNew
        void $ Gtk.printRunPageSetupDialog
          (Just window)
          (Nothing :: Maybe Gtk.PageSetup)
          printSettings

  void $ on pageSetupButton #clicked onClickPageSetup

  #setChild window (Just pageSetupButton)

  window.show

```

The `stack.yaml` is:

```plaintext
snapshot: lts-22.43 # GHC 9.6.6
extra-deps:
- gi-gdk-4.0.9
- gi-gsk-4.0.8
- gi-gtk-4.0.9

```

The `package.yaml` is:

```plaintext
spec-version: 0.36.0
name: gtk-print
version: 0.1.0.0
dependencies:
- base >= 4.7 && < 5
- gi-gtk >= 4.0
- haskell-gi-base
executables:
  gtk-print:
    main: Main.hs
    source-dirs: app
    ghc-options:
    - -threaded
    - -rtsopts
    - -with-rtsopts=-N

```

When the compiled executable is run, it first produces (as expected):

 ![image](https://us1.discourse-cdn.com/flex002/uploads/haskell/original/2X/1/14f43a0485d7da564ec7dc95b8e870d0f46d0ac5.png)

This is the problem, when the ‘Page setup’ button is then clicked, the application aborts with:

```plaintext
(gtk-print.EXE:3184): GLib-CRITICAL **: 18:07:02.332: g_utf8_to_utf16: assertion 'str != NULL' failed

Access violation in generated code when reading 0x0

 Attempting to reconstruct a stack trace...

   Frame Code address
 * 0x9906f49f00 0x7ffd2cf29aba D:\sr\programs\x86_64-windows\msys2-20240727\mingw64\bin\libgtk-4-1.dll+0x2a9aba
 * 0x9906f49f90 0x7ffd2cf2bcba D:\sr\programs\x86_64-windows\msys2-20240727\mingw64\bin\libgtk-4-1.dll+0x2abcba
 * 0x9906f49f98 0x7ff7520a446d D:\Users\mike\Code\GitHub\gtk-print\.stack-work\install\170cbe04\bin\gtk-print.EXE+0x2446d

```

If anybody can show me a similar example which ‘works’ (on any operating system), that might help me narrow what I am experiencing.

---

<div class="post-metadata">

**Author:** ![mpilgrem](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.haskell.org/mpilgrem/32/2529_2.png) [@mpilgrem](https://discourse.haskell.org/u/mpilgrem)\
**Post date:** [December 1, 2024, 6:27pm UTC](https://discourse.haskell.org/t/windows-gtk4-printrunpagesetupdialog-problem/10869/2 "2024-12-01T18:27:27Z")

</div>

Something that, with hindsight, I should have tried first has changed my experience. I’ve updated the Stack-supplied MSYS2:

```nohighlight
stack exec -- pacman -Syu

```

and that updated to `mingw-w64-x86_64-gtk4-4.16.7-1` and `mingw-w64-x86_64-gtksourceview5-5.14.2-1`.

Now, running the same executable still results in the following message being sent to the standard error channel:

```nohighlight
(gtk-print.EXE:46952): GLib-CRITICAL **: 18:23:47.846: g_utf8_to_utf16: assertion 'str != NULL' failed

```

but the Page Setup dialog is presented, and the exectuable otherwise behaves as expected.

So, it looks like it is not the Haskell dimension but a GTK4 bug.