AW: The storage interface cleanup

[Dietmar Maurer <dietmar maurer-it com>]

Dietmar Maurer <dietmar maurer-it com> writes:

> Miguel de Icaza wrote:
> 
> > It seems like a lot of pain inflicted on everyone for the sake of
> > a few changes to Nautilus as Dietmar pointed out.
> 
> I think there is a broad consensus that we need to change the
> Storage interface, and eazel have to cope with that since
> they are using a library (bonobo) that is not even released. Besides,
> none of the eazel people denied that it is not much work to
> make the changes.
> 
> So please let me make the changes. They are necessary and it
> is less work to make them now.
> 

Here is the thing. We were originally supposed to have an API freeze
for Bonobo nearly two months ago, at the start of September. If we
keep breaking the freeze for "would be nice" rather than "must have"
things, GNOME 1.4 will never ship. Every change is potentially
destabilizing, and API changes even more so. For example, there are at
least 100 distinct Nautilus bugs in bugzilla.eazel.com that can be
directly attributed to the Bonobo UI handler change (counting both
fixed and unfixed issues), and that doesn't even count the things that
were fixed without filing a bug. Your change is much smaller than
that, granted, but at some point one needs to draw the line.

[Dietmar Maurer]  I have tried to make the changes before the
freeze. Due to several reasons this was not possible.

I would suggest to take a look at the code that implements
the Storage and Stream interfaces. Many things are broken,
unimplemented or inconsistent (really!!!!!). So I can't take you serious, 
if you say that those changes would destabilize something.
My goal was to fix those bugs, and the first step is the api
cleanup.

If there were some reasonable change review process for Bonobo right
now, I personally would not object to changes like this Stream one -
it seems like it may well be low impact. But as it stands now,
accepting one change feels like opening the floodgates to possibly
many more. 

[Dietmar Maurer]  ??? don't expect an answer.

Personally, I'm gravely disappointed that Miguel and Michael, who
promised us (well, certain Eazel executives) a freeze starting about a
week ago in exchange for us agreeing to port the the new UI handler
and accept it going forward, never properly announced that there was
an API freeze (actually, a semi API freeze, excluding API changes
needed to fix bugs in GNOME 1.4 critical packages, and a few specific
things like the studlyCaps renaming) and have not been enforcing
it. Rather, they expect Eazel people to explain and enforce their
freeze, all the while arguing for more changes.

I don't like being in that position, and I don't think anyone else on
the Nautilus team or at Eazel does either. We're not the Bonobo
maintainers and don't have the authority to make the final decision,
yet we must repeatedly defend the policy that the maintainers promised
us, sometimes against the maintainers themselves.

I feel bad that your changes might not get into Bonobo for GNOME 1.4,
but it's really not the fault of anyone on the Nautilus team that your
patches were ignored for so long. I do think mostly they are good
changes and should have gone in long ago, but I also think it's too
late now.

So we are not going to say no to you, Dietmar. We will leave it up to
Miguel and Michael to keep their own promises.

[Dietmar Maurer]  I think you already agreed to my proposed changes
a week before the freeze (see your old mails). Due to several reasons
I was unable to make the changes before the freeze.

I think it would be nice if we both try to work together. I have already
put some constraints to the api changes. They will have a very minimal
impact on nautilus and those very little changes will prevent us (and
maybe you) from many problems in the future.

Please notice that nautilus is not the only application that
uses bonobo, and many of them use or will use the Stoarge and
Stream interfaces. So we consider this a release critical thing
which will have to be fixed. 

Your argumentation is more or less: ok, your changes make sense, 
but they are not important as nautilus can live with the current api, 
so there is no reason to change it. 
This is IMHO a little bit selfish, considering the fact 
that the required changes for nautilus are small, even if we are
behind the freeze.

So please say "Yes" to the changes.

Regards,

	Dietmar
.