Copy and paste troubles under Artix Linux #46
Comments
|
Do you happen to know if these are pre-existing issues from the old 2.1.2 or 2.0.0 releases? I know Linux builds have never been a well vetted thing. There are multiple PRs and issues on the main repo related to it. The first issue is probably related to the Update checking that happens on startup as it's the only thing I'm aware of making http calls. It causes issues on macOS as well I know. If you change MagicSetEditor2/src/data/settings.cpp Line 185 in 6c5b6a0 If that's the case then it's probably easier to just disable update checking where it actually tries to do it. Either by returning immediately or commenting out the MagicSetEditor2/src/gui/update_checker.cpp Line 129 in 6c5b6a0 I think it's probably safe at this point to just disable/remove the updater entirely. The updates url points at sourceforge which wasn't updated even during that burst of activity in 2021 and still refers to 2.0.0 when the checker runs.. |
|
Ah, I see one of those alluded to PRs if yours. heh |
|
Commenting out the update check does indeed fix the socket issues, but not the format specifier SIGTRAPping. |
|
If I suppress dialog generation when the format specifier error pops up and hit Continue, the last line of the console output is as follows: |
|
I'm a bit confused on why I don't see this call in your backtrace. I'm seeing that in the middle of my call stacks between Does the old 2.1.2 build that you had that PR against run? That'd at least help to narrow it down to changes that I've made. I do have some changes around keycode event handling for unicode text entry, but I wouldn't expect that to crash on load... I'd assume it would wait until text input or something. EDIT: Disregard a bunch of that. MagicSetEditor2/src/gui/welcome_window.cpp Line 126 in 4a60647 |
|
I'll admit that I know nothing about debugging Linux binaries, or indeed binaries on any platform. :( |
|
Here's what I get from doing a gdb |
|
Okay, well there's a failure in there trying to load an icon for https://github.com/haganbmj/MagicSetEditor2/blob/master/resource/win32_res.rc There's a few other new icons in there as well ( The last line referenced is to building up the menus, and that's where it actually fails to find a MagicSetEditor2/src/gui/set/window.cpp Line 501 in 4a60647 |
|
Hm, when I break that icon I do get a similar call stack and a fatal error about loading the resource.
|
|
Yeah, Linux has a Now, that said, when I try to copy and paste a card, I get a new and exciting error: |
halian
changed the title
|
Yay, at least we made some progress. |
|
So looks like your build is using GTK, while the Windows one uses MSW. https://github.com/wxWidgets/wxWidgets/blob/v3.2.0/src/gtk/clipbrd.cpp#L737-L755 If copy/paste is working in other places then it's probably something to do with unmarshaling the clipboard into a custom object? Which is in this block On MSW there's all this checking for known formats and such, but the paste here falls through and uses that There's some interesting behavior that I didn't know MSE did here to put single cards onto the clipboard as images as well. https://github.com/twanvl/MagicSetEditor2/blob/master/src/data/format/clipboard.cpp#L143-L153 Just out of sheer curiosity, if you copy 2 cards does it work since it would bypass that behavior? |
|
So for me on Windows MSE falls through to this call here when getting the In another area there's a mention that implementations should override the method and specify their own, though... So I'm going to take a stab at implementing that and see if maybe that's what's going on here? My expectation would be that this fails for copy/pasting Keywords as well since that's also a custom implementation. EDIT: When you have the opportunity could you try the following change? (b2a94b1) It's on branch |
…#46) Both are wxTextDataObject, which looks to have a default implementation on Windows at least. Unclear how it's resolving on Linux, so hoping that this ensures the behavior.
…#46) Both are wxTextDataObject, which looks to have a default implementation on Windows at least. Unclear how it's resolving on Linux, so hoping that this ensures the behavior.
…haganbmj#46) Both are wxTextDataObject, which looks to have a default implementation on Windows at least. Unclear how it's resolving on Linux, so hoping that this ensures the behavior.
Per topic, when trying to run latest under Artix Linux (Arch derivative), it throws a bunch of socket errors when opening the program, then a bunch of format specifier errors when trying to open a set, the latter of which prove fatal. Here's the console output of one of the former:
and the backtrace of one of the latter:
Additionally, while not show-stopping, scrollable areas (e.g. the new set scene) will refuse to scroll past a point where the selected item is on the edge of the list.
The text was updated successfully, but these errors were encountered: