[Mono-devel-list] possible bug in static constructor evaluation order

[Michal Moskal]

On Mon, May 05, 2003 at 12:56:11PM +0200, Paolo Molaro wrote:
> On 05/04/03 Michal Moskal wrote:
> > Could anyone confirm this with current CVS (I cannot get it to run)?
> 
> Please, can you post details about the issue you have in getting cvs to
> run? Thanks.

I believe I just didn't try hard enough (didn't even uninstall
mono-0.23). I'm developing new language for .net. But I'm now in
mainly-planning stage, so I use mono (and pnet for that matter) just
for running small test cases to learn how the stuff works.

BTW: are there any plans about generics support in mono?

> This is caused by inlining: there are also other issues with static
> initializers ordering both in the current release and in the cvs
> version. They will be addressed after the next release (due real soon
> now) when we'll be able to drop the old jit.
> You'll find, though, that the new jit doesn't use inlining by default,
> so your test will execute correctly with it if you don't explicitly
> enable it.

OK, so it's a bug and I can write in operational semantics, that class
initialization happens just before first use?

-- 
: Michal Moskal :: http://www.kernel.pl/~malekith : GCS {C,UL}++++$ a? !tv
: PLD Linux ::::::::: Wroclaw University, CS Dept : {E-,w}-- {b++,e}>+++ h