[nem-en] Bugs in extension method implementation
Kamil Skalski
kamil.skalski at gmail.com
Sat Aug 26 21:12:54 CEST 2006
Ah, I can see now that you use the same compiler version for ncc.exe
and NC.dll, so this is no version mismatch here. Maybe it is because
the compilation is running from within MSBuild? In such case maybe the
Nemerle.dll loaded by MSBuild is causing the problems (something
similar was with Nemerle.Macros if I remember correctly).
I have an idea - let's remove dependency of MSBuild task on
Nemerle.dll (by not using types from it), maybe this will solve some
problems.
2006/8/26, vc <vc at rsdn.ru>:
> > Indeed, it fails with current boot/ version, but it works correctly
> > with out.stage3/ before your commit.
>
> I not use boot/ncc.exe. I use compiled version of compiler. But I compile
> it using MSBuild (Nemerle.sln).
>
> > Well, this is strange, unless you compile using one compiler, but
> > referencing the other one, like:
> > boot/ncc.exe utility.n -r out.stage3/Nemerle.Compiler
>
> Probably it occurs because I compile a code from VS 2005.
>
> > Could you try again with the version before patch and tell me how
> > exactly do I reproduce this error?
>
> This is compilation log obtained from VS 2005.
> -----------------------------------------------------
> Target CoreCompile:
> C:\Program Files\Nemerle\ncc.exe /no-color /nowarn:1701,1702 /no-stdlib
> /define:DEBUG;TRACE /debug+
> /out:obj\Debug\Ext2.exe
> /target:exe Properties\AssemblyInfo.n Main.n
> /ref:"C:\Program Files\Nemerle\Nemerle.Compiler.dll"
> /ref:"C:\Program Files\Nemerle\Nemerle.dll"
> /ref:C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727\System.Data.dll
> /ref:C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727\System.dll
> /ref:C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727\System.Xml.dll
>
> D:\MyProjects\Tests\Nemerle\Ext2\Ext2\Main.n(5,14,5,28): error : there is
> no member named `GetName' in Nemerle.Compiler.IMember with type ?
> -----------------------------------------------------
> C:\Program Files\Nemerle\ contains compilers binaries r6576 (compiled
> from source, not boot).
>
> > I don't like the patch, since IsDefined against assembly loaded into
> > SystemTypeCache is the *correct* (and probably fastest) way to check
> > for attributes.
>
> In my option ExtensionAttribute not usual attribute. As to speed... I do
> not think, that it will affect performance. Not many of classes containing
> extension methods. Against a background of time of loading of metadata it's
> are nothing.
>
>
>
>
> _______________________________________________
> https://nemerle.org/mailman/listinfo/devel-en
>
--
Kamil Skalski
http://nazgul.omega.pl
More information about the devel-en
mailing list