[Mono-devel-list] Re: System.XML warning

[Kelly Leahy]

Hi all...

While I haven't looked at the code this refers to
myself, it seems to me that a lot of these warnings
are probably related to private functions put in place
to support public functionality that is not yet
implemented.  Is this the case?

Marek, while we are adding new compiler warnings, can
we have some way to "hide" the warnings on an
individual symbol basis?  It seems to me that our goal
should be to get rid of all warnings one way or
another.  If the warning is coming up because we
haven't used a function we know we will use later,
it'd be nice to attach an attribute to that function,
for instance, that tells our compiler not to show a
warning when this function is private and unused.  If,
on the other hand, we messed up and just ended up
having a private symbol that we wont use, it'd be nice
to get a message on that one.

It seems to me that just turning off the warnings
doesn't help much in these "partial development"
situations.

What do you guys think?

Kelly 

--- Atsushi Eno <atsushi at ximian.com> wrote:

> Hi Marek,
> 
> Oh, what you pointed out is really important.
> Because if all
> the contributors don't want to fix any of assemblies
> because of
> the number of warnings, it is so important matter
> for all the
> hackers and thus we can let people hacking on mono.
> 
> OK, I'll look into all the warnings.
> 
> Atsushi Eno
> 
> Marek Safar wrote:
> > Hello Eno,
> > 
> >>
> >> Can you post it to public devel-list? Because
> 1)Lluis would also
> >> be interested in it, for sys.xml.serialization,
> and 2)if there
> >> was such warning messages that does not make
> sense that would
> >> also apply to other assemblies (I think you must
> have sent similar
> >> messages to each developers), it is waste of time
> that only you
> >> and I discuss them.
> > 
> > 
> > Not exactly, I have problem with so many warnings
> only for System.XML 
> > and System.Web assemblies.
> > My tactic is to address you as main developer of
> System.XML to reduce as 
> > many as possible warnings.
> > But it doesn't mean ALL, just reduce the number of
> warnings to 
> > acceptable level (let say up to 20) and then other
> people can
> > contribute later to fix the rest of them.
> > Simply I would like to avoid warning flooding
> where we "unreasonably" 
> > reports 60 or more warnings for one assembly.
> > Many people get confused from this and mainly they
> don't want to start 
> > fixing where warning output is so big.
> > 
> > Marek
> > 
> >>> Hello Eno,
> >>>
> >>> I would like to commit my next mcs patche which
> improves our warning 
> >>> detection.
> >>> However, same as in the first case I would like
> to reduce the number 
> >>> of BCL warnings to acceptable level.
> >>> I am going to enable warning CS0169 and in the
> attachment is list of 
> >>> warnings in System.XML namespace compiled with
> >>> improved mcs. Could you please have a look at
> this list and either 
> >>> remove them or give me hint which of them can be
> removed (I expect 
> >>> almost every).
> >>>
> >>> Please be noticed that this is output of mcs
> compiler and any other 
> >>> compiler is not able to detect this warning as
> good as mcs does.
> >>>
> >>> Regards,
> >>> Marek
> >>>
> >>>