On Sun, 10 Jan 2010, Andreas Raab wrote:

> David T. Lewis wrote:
>> On Sun, Jan 10, 2010 at 03:09:10AM +0100, Levente Uzonyi wrote:
>>> Why aren't these methods in the 311Deprecated package?
>> 
>> Good question, and I'm not sure which is better in this case. I think
>> the choice should be driven by whether we view these as deprecated,
>> implying that they would be removed perhaps a year or so from now, or
>> if they are intended as compatability methods that might stay in the
>> image for a longer period of time.
>
> I agree. The reason why I put findElementOrNil: back right away was that it's 
> being used by DynamicBindings of which a multitude of versions exists in many 
> places on the web. It's virtually impossible to fix all the existing 
> versions.
>
>> I'm pretty sure that the folks doing e.g. Seaside are going to feel
>> the same way, so we need to be careful to provide compatibility
>> methods where appropriate, but not go overboard with it and clutter
>> up the image with junk. So, to return to the original question,
>> I'm not sure if these methods should be moved to 311Deprecated,
>> but if the intent is to keep them for say five years for compatibility,
>> then no they should not be moved.
>
> +1 on all of it. We should get rid of these methods whenever we can but we 
> also need to take into account the existence of external packages. In the 
> case of findElementOrNil: I first found (and fixed) a usage in Tweak, then I 
> found (and fixed) a usage in DynamicBindings on http://source.squeak.org/ss 
> but then I ran into one usage via DynamicBindings in a package loaded three 
> indirections deep via Metacello. That's what pushed it over the edge for me - 
> clearly there's more than one version out there and it's hard to predict 
> which one people will attempt to load.

There's a DynamicBindings repositiory at squeaksource.com 
(http://squeaksource.com/DynamicBindings.html ) which seems to be 
abandoned. AFAIK 
the package is maintained at http://squeaksource.com/KomHttpServer.html .


Levente

>
> Cheers,
>  - Andreas
>
>