Re: [boost] Re: Re: Mini-reviews

From: Pavol Droba (droba_at_[hidden])
Date: 2004-07-29 03:23:16

Hi,

On Wed, Jul 28, 2004 at 03:47:14AM +1000, Thorsten Ottosen wrote:
> 
> 
> "Doug Gregor" <dgregor_at_[hidden]> wrote in message news:7DEAB5E4-DFD3-11D8-BD44-000D932B7224_at_cs.indiana.edu...
> |
> | On Jul 27, 2004, at 6:42 AM, Thorsten Ottosen wrote:
> 
> | > contribution will fit into the overall scheme. Then a formal
> | > mini-review should follow.
> |
> | At what point are there enough algorithms under the same category that
> | we should just call it a "full" review?
> 
> Good question! I don't gave an definite answer.
> 
>  I would prefer that the main contribution is organized by a few people; this main contribution
> should then be given a full review. And then
> extra small contributions are mini-reviewed. Take the string-algorithms as an example. I hope Pavol will encourage people
> to add extra functions and work out their interface with him and others on the list.
> 
> I don't see the first real review happening without some group with the main responsibility. And I don't see very small
> contributions happening on their own because I fear the big picture is lost.
> 

I completely agree. Mini-reviews are very good idea, provided, the there is a person/group that is responsible
for the overal picture for a particular algorithms group.

During the development of the string algo lib, I have found that there are several useful facilities that
are orthogonal to an algorithm functionality. A good example is the facility that evolved to Boost Range library.
These facilities should be reused by all algorithm. 
In addtion issues like directory organization, namespace usage and etc. are also important to form a consistent
algorithmic library.

If core issues are settled, than the "mini-review" approach can be very good alternative.

Regards,

Pavol.