[Squeakfoundation]Brainstormin'

[ducasse]

on 29/01/02 9:12 PM, Tim Rowledge at tim@sumeru.stanford.edu wrote:

> Mark Guzdial <guzdial@cc.gatech.edu> is claimed by the authorities to have
> written:
> 
>> Bug fixes are easy -- just harvest them and patch them in.
> Even tiny bug fixes need to be carefully considered. Oft times somebody
> will find something that seems not to work for them and 'fix' it; almost
> as often it turns out to completely screw up some other pat of the
> system. You need to know a _lot_ to do the job properly.
> 
>> Enhancements, especially big ones, need to be made carefully.  I hope
>> that as decisions get made to put new things in, the decisions are
>> made while considering whether we can afford to KEEP the new things
>> in, i.e., is the developer promising to stick around and keep it
>> going.  I don't mean to slight the developers of Linda, Connectors,
>> etc.  I just think that maintaining the enhancement belongs on the
>> decision criteria list.
> Exactly. This is one of the reasons why I always push for
> a) reviews
> b) "If it isn't documented, it doesn't exist. If it doesn't exist, how
> can we incorporate it?"

That's why if people can at least define tests that covers what they fixed
this helps a lot the process of harvesting. Even if they may break something
else in another place. Building tests is the way to go.

For example, I was harvesting the fix of hans-martin on big numbers hexa
stuff. I cannot access that the code is good because I'm bad with this kind
of maths but I can access that if you do 100 bitShiftLeft then 100
bitShiftRight you should obtain the same number. ;)

So tests tests tests.....

amen.