Re: Major enhancements to pearweb

Lukas Kahwe Smith wrote: > Greg Beaver wrote: > >> Note that the age of bugs is the important thing here. For instance, >> the Pager package is ranked 169 out of 170 packages with open bugs >> (bad), even though it has only 1 open bug, because the bug is more than >> 2 years old. This bug could easily be resolved (excellent solution is >> suggested by the original opener, Alan Knowles), and is an example of a >> bug that should be cleaned up as it helps to give PEAR a bad name for no >> good reason at all. By fixing and closing this bug, Pager would go from >> #169 to #1 instantly. Everyone wins. > > I do not want to make this all too complex, but it might make sense to > also cram in there some information about the number of downloads for > the release with the highest stability level and some relation to LOC > as well. But generally I think its an awesome addition. > > The important point of this is that it should not discourage > developers because they feel like their package is ethernally branded > as low quality. So there has to be a clear chance to improve the > ranking. This is especially important if maintaince of a package > changes. As a result I think the history should only cover the last 12 > months. Hi Lukas, I implemented, and then commented out, the possibility of doing exactly this. However, I thought a more representative metric was the *frequency* of downloads, rather than the number. This will allow an exciting new package to jump to #1 even though it has only 10,000 downloads, because these downloads happened over 16 hours, as opposed to something like XML_RPC which has a gazillion downloads. Once we get pearweb synced from the pearweb package, these innovations can be tested, benchmarked, etc. Greg