Migrate to bindbc #612
Migrate to bindbc #612
Conversation
|
I've updated previous work and merged with master, tested on Windows and Ubuntu 22.04 (both X11 and SDL2 backends). Should be good to go now, please merge. |
| @@ -11,7 +11,7 @@ | |||
| "targetName": "dminer", | |||
| "targetType": "executable", | |||
|
|
|||
| "sourceFiles-windows": ["$PACKAGE_DIR/src/win_app.def"], | |||
| "sourceFiles-windows-x86": ["$PACKAGE_DIR/src/win_app.def"], | |||
There was a problem hiding this comment.
x86 arch uses old DMD linker (do not confuse it with x86_mscoff), Microsoft linker doesn't understands this and emits warning, but instead of completely removing this let's put it where it belongs to.
examples/opengl/src/openglexample.d
Outdated
| @@ -136,7 +135,8 @@ class MyOpenglWidget : VerticalLayout { | |||
| return; | |||
| } | |||
| bool canUseOldApi = !!glLightfv; | |||
| bool canUseNewApi = !glSupport.legacyMode; | |||
| //bool canUseNewApi = !glSupport.legacyMode; | |||
There was a problem hiding this comment.
Commented code should be removed or uncommented
src/dlangui/core/editable.d
Outdated
| @@ -24,6 +24,7 @@ import dlangui.core.linestream; | |||
| import dlangui.core.streams; | |||
| import std.algorithm; | |||
| import std.conv : to; | |||
| import std.uni; | |||
There was a problem hiding this comment.
Should this modules be imported at all?
There was a problem hiding this comment.
Yes, can be made static though.
dlangui/src/dlangui/core/editable.d
Line 1211 in 1b77b9b
src\dlangui\core\editable.d(1211,28): Deprecation: package std.uni is not accessible here, perhaps add 'static import std.uni;'
src/dlangui/platforms/sdl/sdlapp.d
Outdated
| // NOTE: this has crappy complexity as it was just updated as is | ||
| // it also does not checks if the symbol was actually loaded | ||
| auto errMsg = cast(string) info.message[0 .. info.message.strlen]; | ||
| bool found = names |
There was a problem hiding this comment.
I guess this code is wrong because it finds errMsg in names instead of names in errMsg. Also using array to count items is a bad practice.
| bool found = names | |
| bool found = any!(s=>errMsg1.canFind(s))(names); |
There was a problem hiding this comment.
that's why there is a note, it is definitely ugly but this code is basically just run once on startup, it doesn't do much harm, so it can be improved later
There was a problem hiding this comment.
It's not just ugly, it's wrong. Considering you add this code it should be fixed in this PR, not later.
There was a problem hiding this comment.
Yep, I agree it sucks, but the overall logic still valid, the problem is that it searches for substring in array of strings - it doesn't matter whether it searches name in errors or errors for name, in any case it is MxN task, but indeed as it was updated just as is there also another loop which makes it even worse.
It will throw exception if some symbols were unable to be loaded, another option would be to simply log an error and let it crash on null pointer call at some stage later which I think is far more frustrating to both programmer and a user.
I'll see what I can do to improve it.
There was a problem hiding this comment.
Well, as I understand it does matter whether it searches name in errors or errors for name - definitely you will always fail if you try to find the whole error message in a set of function names just because the error message is longer than some function name and contain other words than a function name.
There was a problem hiding this comment.
Oops. Indeed that was my overlook, looking at phobos unit tests I misinterpret that canFind will pick up substring as well, but what's really needed is indexOf from std.string
src/dlangui/graphics/glsupport.d
Outdated
| // NOTE: this has crappy complexity as it was just updated as is | ||
| // it also does not checks if the symbol was actually loaded | ||
| auto errMsg = cast(string) info.message[0 .. info.message.strlen]; | ||
| bool found = names |
There was a problem hiding this comment.
I guess this code is wrong because it finds errMsg in names instead of names in errMsg. Also using array to count items is a bad practice.
| bool found = names | |
| bool found = any!(s=>errMsg1.canFind(s))(names); |
| // NOTE: this has crappy complexity as it was just updated as is | ||
| // it also does not checks if the symbol was actually loaded | ||
| auto errMsg = cast(string) info.message[0 .. info.message.strlen]; | ||
| bool found = names |
There was a problem hiding this comment.
I guess this code is wrong because it finds errMsg in names instead of names in errMsg. Also using array to count items is a bad practice.
| bool found = names | |
| bool found = any!(s=>errMsg1.canFind(s))(names); |
|
Can you please rebase your changes on master. instead of merging with it? I'm afraid we will have a lot of garbage in git history if it's not rebased. |
Superbelko
force-pushed
the
migrate_bindbc
branch
from
1b77b9b
to
a65085d
Compare
|
I've fixed error checking issue and squished the commit, should be ok now. |
|
Thank you for your contribution! |
BindBC is a newer loader mechanism that replaces Derelict, as Derelict is now basically discontinued I've migrated dlangui to use bindbc instead.
Note: provided .dll/.so needs to be updated as well, otherwise it will fail to load as the provided FreeType and SDL2 libs are too old