[Mono-dev] [patch] coreclr/com check + icall cleanup

Rodrigo Kumpera kumpera at gmail.com
Thu Nov 5 15:55:55 EST 2009


The new patch looks good.


On Thu, Nov 5, 2009 at 6:30 PM, Sebastien Pouliot <sebastien at ximian.com>wrote:

> On Wed, 2009-11-04 at 16:31 -0200, Rodrigo Kumpera wrote:
> > The icall removal patch is ok.
>
> committed
>
> > The second one is tricky. Do we really want to completely disable COM
> > support when running under the sandbox?
>
> I don't see this as an immediate issue but...
>
> > It does make sense for moonlight, but not for other users of coreclr.
> >
> > I believe we should only fail COM for non-platform assemblies which
> > has the same result for moonlight
>
> a new patch is attached.
>
> > but won't
> > bite future users of the sandbox code.
>
> Well it won't change anything for Moonlight[1] but it will still bite
> any other (well future) user of coreclr unless the BCL they provide
> offers the required COM types [2]. Otherwise it will simply abort (like
> id does today).
>
> Sebastien
>
> [1] unless someone adds a [ComImport] somewhere in the platform code -
> but that would not pass our test suite :)
>
> [2] A added a FIXME in the patch about this. In any case the g_abort
> should make it clear enough to runtime embedders
>
> >
> >
> > On Thu, Oct 29, 2009 at 4:43 PM, Sebastien Pouliot
> > <sebastien at ximian.com> wrote:
> >         Hello,
> >
> >         Two small/easy patches for review.
> >
> >         The first one avoid calling mono_com_init when coreclr is
> >         enabled*.
> >         This avoid a crash if some assembly use [ComImport] on a type
> >         and throw
> >         a TypeLoadException - which is what happens in Silverlight.
> >
> >                * For some reason (I guess it use COM for it's platform
> >         code,
> >                while Moonlight does not) Silverlight expose
> >         [ComImport] but
> >                otherwise does not support COM (as least for
> >         application code).
> >
> >         Second patch removes some internal calls (all strings except
> >         one) that
> >         are not used (anymore) in the class libraries.
> >
> >         Sebastien
> >
> >         p.s. both patches were created from 2-6 branch but I'll commit
> >         them
> >         against HEAD too.
> >
> >         _______________________________________________
> >         Mono-devel-list mailing list
> >         Mono-devel-list at lists.ximian.com
> >         http://lists.ximian.com/mailman/listinfo/mono-devel-list
> >
> >
> > _______________________________________________
> > Mono-devel-list mailing list
> > Mono-devel-list at lists.ximian.com
> > http://lists.ximian.com/mailman/listinfo/mono-devel-list
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20091105/a3db393f/attachment.html 


More information about the Mono-devel-list mailing list