Re: PFC-RFC

On Monday, February 24, 2003, at 04:40 PM, Christian Stocker wrote:

On Mon, 24 Feb 2003, George Schlossnagle wrote:

I'd be concerned about a prevalence of me-too-ism. I.e. how can I say no to Stig even if his code sucks (I like Stigs code, but he's a prominent figure easy to pick on).

peer-reviewing is in my oponion not about "your code sucks", everyone can do that, but more about "hey, why did you do that this way and not that way". It's more about improving the code and not about deciding, if it's good enough. I know, this is quite an ideal-world assumption, but maybe worth the effort...

I agree on peer-reviewing, but I see it as hard to achieve in a all-volunteer community project like this. I can't count the number of projects I've +1'd because they sounded like nice ideas. If I had to actually review it, would I? If the pressure was on and it was a friends project that was import to make a PFC (possibly because something else depending on it was inline to be a PFC), would that sway my vote and make me inclined to just give it a thumbs up instead of rigorously auditing the code?

I personally dig a community rating system. If 500 people think that package foo is the greatest thing since sliced cheese, then I don't really care that it is a PFC, it probably merits consideration.

My proposal doesn't prevent that You still can download whatever class is in PEAR, a PFC does not force you to use that. I just don't want to end up in CPAN-land, where I have no idea, which class is the "best" one.

My point was that an open ranking system actually gives you an idea of which is the most relevant. If there is a class whose description matches my requiremnts but which has 2 stars and 10 votes, I know I'm probably off just starting from scratch. if it has 10 votes with an average of 4 stars, I know it's worth looking at. I think by opening the rating system like this you get less urbane reviews and more a feeling of how useful.well-implemented the class is from a real world perspective. I also think that an open revieiwing system distributes the burden over the entire user population and not just on the shoulders of key developers who probably have other things they would like to be doing with their lives.