[DotGNU]Re: [Pnet-developers] Re: back to Emit problems

[Alex Stephens]

On Sat, 1 May 2004 15:52:30 +0200, Jonas Printzen wrote:

> Hi Gopal!
> 
> My progress is a bit slow. Mainly because I have failed
> to free more than an hour now and then for this.
> This will improve...
> 
> Found the part in the code that trows the exception
> when invoking a dynamically generated entry.
> 
> Looking for traces of debugging support in the code.
> 
> Trying to get a grip of execution from two perspectives, 
> runtime loaded assembly and generated. There are obviously
> some difference that need to be removed. 
> 
> I assumed I would find a common base for generating
> IL between compiler and Emit. Not so I think. Design issue?
> 
> I had the same thought, unit-tests are a good start. Begun to
> look around for what exists and what shape last night but
> was interrupted. This is my next step.
> (I never change code without some kind of regression protrection)
> 
> Was laso looking for a way to extend the exceptions
> thrown from engine/convert.c with a string or so.
> Just to get more info on why something failed.
> There are 4 paths in the code thet generates the
> EntryPointNotFoundException. Need to pin it down
> to not waste time on wrong path. And expect the same
> furter down the line..
> 
> What is your typical approch? gdb, traces
> code-reading and neural execution? :o)

That's a good question. Is there any way to debug programs created with pnet? I 
mean some sort of gdb-like program.

Alex Stephens

p.s. I like the term "neural execution", sounds better than code-walkthrough or 
something like that.