[Mono-list] self-hosting assemblies status update

[Martin Baulig]

Hi,

the latest status is now:

=====
[martin@einstein corlib]$ mono --debug ~/monocvs/mcs/mcs/mcs-mono2.exe --fatal @unix.args

(process:29065): ** WARNING **: unhandled exception System.NullReferenceException: "A null value was found where an object
instance was required"
in [0x000a6] (at corlib.il:165013) System.Reflection.Emit.TypeBuilder:SetCustomAttribute (System.Reflection.Emit.CustomAttributeBuilder)
in [0x00386] (at attribute.cs:626) Mono.CSharp.Attribute:ApplyAttributes (Mono.CSharp.EmitContext,object,object,Mono.CSharp.Attributes,Mono.CSharp.Location)
in [0x00336] (at class.cs:1350) Mono.CSharp.TypeContainer:Emit ()
in [0x001aa] (at rootcontext.cs:694) Mono.CSharp.RootContext:EmitCode ()
in [0x00a13] (at driver.cs:853) Mono.CSharp.Driver:MainDriver (string[])
in [0x00002] (at driver.cs:198) Mono.CSharp.Driver:Main (string[])

RESULT: -1
=====

I can probably look at this on Saturday.

> Anyway, we made quite a progress in the last few days!
> We should be able to start building the assemblies daily to catch
> regressions and maybe also start running the test suite against them.

Well, running the test suite doesn't help finding these bugs.

Most of the problems which are currently preventing us from compiling corlib
only exist when compiling corlib, so you won't catch them by running the
test suite.  Most of these bugs are type mismatches between a system type
and the TypeBuilder of that type - hard to find if you don't know where you
need to look, but most of them should now be fixed anyways.

However, we should start running the test suite anyways to make sure we do
not break mcs while trying to make it compile corlib.  At the moment, all
135 tests are working fine on Linux.

-- 
Martin Baulig
martin@gnome.org