[Bese-devel] request dispatching

[Friedrich Dominicus]

Wojciech Kaczmarek <wojtekk at kofeina.net> writes:

> On 14 Mar 2006, at 9:44, Friedrich Dominicus wrote:
>
>> "Attila Lendvai" <attila.lendvai at gmail.com> writes:
>>
>>> Wouldn't that mean that this breaks every existing application?i'm a
>>> newcomer to lisp, and one of the most positive aspects of this
>>> community is that people tend to not care that much about backward
>>> compatibility... i understand that this may be an issue with deployed
>>> instances, etc., but there are many ways to handle that.just my
>>> 0.02, - attila(alias 101 on irc &no 'its not lisp code :)
>> I disagree, backward compatability is not just a game. It's crucial
>> for any kind of "serious" investment. Even M$ has gone to extremes to
>> keep backward compatbility. As you can see in there income that was a
>> wise decision. Now they are drifting away and suddenly people are not
>> willing to follow without 'thinking'.
>>
>> Intel has learned it the hard way also, since they decided to build
>> the I64 stuff, AMD has taken away a lot of market share.
>>
>> The tendency at the moment in the GNU world is unfortunatly, "breaks
>> as much as you can", how many packages still use GTK 1.2 because they
>> "decided" it's a right thing to break everything.
>>
>> I like Common Lisp because it is quite stable and I can take code from
>> back a few years and have a good chance getting it running even in the
>> newer versions. This is also another big plus for C instead of C++.
>>
>> you just can think that backward-compatiblity is for weenies if you do
>> not have large applications. Otherwise you would not write such stuff.
>
> I think you can find (and meet) a significant number of people
> claiming that backward compatibility 'myth' is evil in industry and
> some compatibility practices are leading to a least-common-denominator
> effect, which often kills innovations. 
Yes that is the case, but you have to consider also the costs. Of
course if you get paid for it to follow all the "corrections" this
might be acceptable although I doubt that anyone will be very happy to
rewrite code over and over again, introducing new bugs on the way and
not getting anything new done. 

But I see your arguments also, but I feel you should not take it all
to light, software has a unbeliavable resistance. And things supposed
not to survive over a few years are still hanging around. 

> I'm not willing to start a
> flame here; please just consider that, in the case of our UCW
> dispatcher changes, one can learn a lot. Fortunately Marco seems keep
> macros for old behaviour so we aren't forced to learn it all right
> now. Best of both worlds.
I can't see you flaming and I can't see me flaming also. We have
different opinions and we have to accept it. I just think just for
getting something bette done kiling every application out there is not
a wise choice. I'm quite happy if the defentry stuff will stay. I can
see the points of collecting things in one place but I can see also
see the point keepin the defentry near the stuff which is behind it.

So I think it is the "right" thing what Marco has done. 

Regards
Friedrich