Re: [Haskell-cafe] real haskell difficulties (at least for me)

On Thu, 2009-01-15 at 14:25 +0100, Ketil Malde wrote:

Duncan Coutts writes:

...

On Thu, 2009-01-15 at 09:49 +0300, Eugene Kirpichov wrote:

...

Would be nice if after a failed build cabal asked whether or not to upload its log immediately, and (on the hackage side) this led to an email being sent to the maintainer.

...

It should not be quite that synchronous but yes that's the general idea.

Perhaps the maintainer should receive a build summary at regular intervals? This would also work as a check whether there is a live maintainer at the other end of the listed maintainer address. (And hopefully be enough of an annoyance on the libraries@ list that people would start looking for maintainers for orphaned packages :-)

It may well be tempting to plague maintainers until they fix their packages however in practise it will not work. We want a low barrier to entry for packages on hackage and we do not want to annoy package maintainers to the point where they decide to stop using hackage at all. Of course we also want to measure package quality. The approach I've always advocated is to define subsets of hackage that do meet higher quality levels. Be defining such subsets and holding them up as being a "good thing" in the community then it puts social pressure on maintainers to have their packages in those higher quality subsets. Then to help them do that they may choose to opt-in to various services like receiving a build summary at regular intervals. So it's the same goal but without the stick. Duncan