On 5/13/20 1:56 PM, Rafael Fonseca wrote:
This patch series convert various simple instances of virObject to a GObject equivalent.
I'm sorry upfront for misusing your patchset and I'm also sorry for bringing this up again.
I think we need to step back and re-evaluate whether this is worth it. GObject is horrible and frightening part of GLib. Not only one has to define empty functions, but we can't mix virObject and GObject. For instance: virObjectRef() called over GObject will corrupt memory.
Worse, there is no way to check whether your patches converted everything (and clearly they did not because I see couple of tests crashing; or maybe they did at the time of send but now the code has changed - anyway, point proven).
I started reviewing and stopped at the first patch realizing, I have no idea whether you converted every virObjectRef()/virObjectUnref() with corresponding glib call.
I also wanted to write a cocci spatch that might at least identify places where that is happening, but apparently my spatch skills are poor.
Does anybody have an idea how to verify these patches? Michal
