Martin Wehking

> > can be run independently from the already existing apron analyses. > > > Abstract states in the newly added domain are represented by structs containing a matrix and...

> > Currently, it's for example not possible to run the polyhedra and affeq analysis simultaneously. That's what we wanted to change. > > Currently (on `master`) there's no affeq...

> This is a huge diff (by necessity) and that makes it very hard to review. Could you maybe provide an overview over the different modules (and module types) that...

> > * Another major change is that functions with side effects (usually ending with "_with") defined inside `apronDomain` were replaced by functional equivalents > > This point is easily...

> > * implementations for vectors and matrices (realized through lists and lists of lists. Could be replaced in future > > This is also a major point when it...

> Might make sense to also add a regression test for such things, such that we'll notice if we inadvertently break it again in the future! 56b4125b384ef9a89f249f65e643ba036153a236 I've manually reduced...

> Why does this PR update `g2html`? Were there any necessary changes, or is this by mistake? I've reverted the updates to all submodules. These were indeed by mistake 54907736c4d93fd9e4e35d2870839d394ef0f30a...

I agree with @ldrumm and @jchlanda. It would definitely be better to not have 2 different styles for unified runtime and intel/llvm.

> @MartinWehking, why this change is done only for NVPTX target? Can't we make the same change for other targets as well? Hi @bader We also have a patch for...

@jinge90 That sounds like a reasonable concern. If there are no objections I would leave it as it is for this patch.