Page 4 of 5

Re: 11.0 bugs

Posted: Tue Apr 10, 2018 12:01 am
by slime
pgimeno wrote:
Mon Apr 09, 2018 11:18 pm
Thanks for the clarification. If PhysFS provides a method to check the destination of a symlink, I'd appreciate if a future version exposes it.
It doesn't (yet), unfortunately.

Re: 11.0 bugs

Posted: Tue Apr 10, 2018 11:23 am
by fendrin
love.filesystem.getDirectoryItems( dir )
with dir being a symlink to a directory returns an empty table.

love.filesystem.areSymlinksEnabled( ) is true.

This pretty much breaks symlink support for my purposes, haven't found a workaround yet.

Re: 11.0 bugs

Posted: Tue Apr 10, 2018 6:32 pm
by bartbes
Even if you append a slash?

Re: 11.0 bugs

Posted: Wed Apr 11, 2018 3:16 am
by fendrin
bartbes wrote:
Tue Apr 10, 2018 6:32 pm
Even if you append a slash?
Sadly, still an empty table.

Re: 11.0 bugs

Posted: Wed Apr 18, 2018 3:24 am
by drunken_munki
In both 11.0 hotfix and 11.1 for me calling love.window.setMode() seems to crash all spritebatches and they no longer can be drawn into or drawn onto screen, just black output similar to the previous canvas related bug in original 11.0.

Anyone else seen this issue?

I've tried making my call to change the mode synced either before or after both update and draw routines and tested each case, always the same thing occurs.

Have had a deeper look, it seems that the two spritebatches I used get wiped each time love.window.setMode() is called.
Is this expected? If I rebuild them then everything works fine, thankfully!

Re: 11.0 bugs

Posted: Sat Apr 28, 2018 9:46 am
by pgimeno
In a shader, TransformMatrix does not seem to work any longer. It seems to be always a unit matrix.

Was it dropped? If so, this would be a documentation problem.

Re: 11.0 bugs

Posted: Sat Apr 28, 2018 3:12 pm
by slime
Automatically batched vertices are transformed on the CPU, so TransformMatrix doesn't include the current love.graphics transform in auto-batched draws.

Re: 11.0 bugs

Posted: Sat Apr 28, 2018 3:30 pm
by pgimeno
Dang, thank you. I've added that to the documentation.

Re: 11.0 bugs

Posted: Sat May 12, 2018 7:00 pm
by pgimeno
pgimeno wrote:
Wed Apr 04, 2018 4:53 pm
To hopefully prevent it from getting lost in the other thread, and since I can't use the issue tracker submit form in BitBucket, I'll add it here too: when building as static, pthread is missing from the link line.
I've looked into this. Apparently, commit 3086:a0827b5e39b7 introduced a dependency on pthread, yet that dependency is not reflected in the build system. Somehow, dynamic builds get along with it, but static builds fail.

After a few hours of struggling with cmake, I noticed that that's the wrong tool in this case, so after a few more hours of struggling with autoconf, I came up with two possible solutions, the easy, quick'n'dirty one and the complex, do-the-right-thing one.

The easy one is:

Code: Select all

diff -r 0d1c0ea65da0 platform/unix/genmodules
--- a/platform/unix/genmodules  Sun Dec 10 00:16:13 2017 -0400
+++ b/platform/unix/genmodules  Sat May 12 20:53:39 2018 +0200
@@ -149,7 +149,7 @@
 liblove${love_amsuffix}_la_LIBADD = \
        \$(SDL_LIBS) \$(freetype2_LIBS) \$(lua_LIBS)\
        \$(openal_LIBS) \$(zlib_LIBS) \$(libmodplug_LIBS)\
-       \$(vorbisfile_LIBS) \$(theora_LIBS)
+       \$(vorbisfile_LIBS) \$(theora_LIBS) -lpthread
 
 EOF
 
The complex one requires installing ax_pthread.m4 from here: http://git.savannah.gnu.org/gitweb/?p=a ... pthread.m4 to platform/unix/m4 and then applying this patch:

Code: Select all

diff -r 0d1c0ea65da0 platform/unix/configure.ac
--- a/platform/unix/configure.ac	Sun Dec 10 00:16:13 2017 -0400
+++ b/platform/unix/configure.ac	Sat May 12 20:56:18 2018 +0200
@@ -63,6 +63,7 @@
 ACLOVE_DEP_SDL2
 ACLOVE_DEP_LIBM
 ACLOVE_DEP_ZLIB
+ACLOVE_DEP_PTHREAD
 
 # Conditional dependencies
 AS_VAR_IF([enable_module_audio], [yes], [ACLOVE_DEP_OPENAL], [])
diff -r 0d1c0ea65da0 platform/unix/deps.m4
--- a/platform/unix/deps.m4	Sun Dec 10 00:16:13 2017 -0400
+++ b/platform/unix/deps.m4	Sat May 12 20:56:18 2018 +0200
@@ -19,6 +19,9 @@
 AC_DEFUN([ACLOVE_DEP_LIBM], [
 	AC_SEARCH_LIBS([sqrt], [m], [], [LOVE_MSG_ERROR([the C math library])])])
 
+AC_DEFUN([ACLOVE_DEP_PTHREAD], [
+	AX_PTHREAD([], [LOVE_MSG_ERROR([libpthread])])])
+
 AC_DEFUN([ACLOVE_DEP_PHYSFS], [
 	AC_SEARCH_LIBS([PHYSFS_init], [physfs], [], [LOVE_MSG_ERROR([PhysicsFS])])])
 
diff -r 0d1c0ea65da0 platform/unix/genmodules
--- a/platform/unix/genmodules	Sun Dec 10 00:16:13 2017 -0400
+++ b/platform/unix/genmodules	Sat May 12 20:56:18 2018 +0200
@@ -149,7 +149,7 @@
 liblove${love_amsuffix}_la_LIBADD = \
 	\$(SDL_LIBS) \$(freetype2_LIBS) \$(lua_LIBS)\
 	\$(openal_LIBS) \$(zlib_LIBS) \$(libmodplug_LIBS)\
-	\$(vorbisfile_LIBS) \$(theora_LIBS)
+	\$(vorbisfile_LIBS) \$(theora_LIBS) \$(PTHREAD_CFLAGS) \$(PTHREAD_LIBS)
 
 EOF
 
(edited to fix the error message if pthread is not found)

Re: 11.0 bugs

Posted: Sat May 12, 2018 7:46 pm
by bartbes
That's interesting because some dependency (I think SDL2) adds the -pthread flag for me.