So I sort of got past this problem, but one difference I noticed between gmcs and .net 2.0's csc is that in ms if you explicity constrain a generic definition with the new() constraint, the type constraint can be abstract / otherwise not explicitly implement a public, parameterless constructor.
But in gmcs, you have to, which is a bit unfortunate, because I'd like to be able to constrain my generic to a subset of types, but the base class of the types needs to be abstract. seems like a pretty typical scenario.
The other issue I noticed that is different involves type conversions. Gmcs complains that I cannot convert from type X to T, even if type X derives from T. I can cast it to an object, and then to a T -- so there is a workaround, but i'd rather not.
Am I misusing generics? I'm new to them, never used templates in C++ (at least, never wrote any), so I'm not totally sure about the design principles that apply here.
cheers,
--ben joldersma
On 3/22/06, Julien Sobrier <[EMAIL PROTECTED]> wrote:
Ben Joldersma wrote:
> Hello all, I'm trying to compile my .net 2.0 application to mono
> 1.1.13.4 <http://1.1.13.4>. I'm getting the following error from nant
> (I'm using the latest binary.) My project builds corrrectly with .net
> 2.0. There are several places in my project were I require the new()
> constraint. Any suggestions would be much appreciated!
>
> --ben
Hello,
I was about to ask the same question: some of my classes have an implied
constructor MyClass(), but it looks like I have to include this
explicitly with gmcs, otherwise I get the error
WARNING **: Missing method .ctor in assembly ... a run time
I'll look deeper into this.
Julien
--
--ben joldersma
http://ben.creationsnetwork.org
_______________________________________________ Mono-list maillist - [email protected] http://lists.ximian.com/mailman/listinfo/mono-list
