On Thu, Aug 07, 2003 at 04:51:15AM -0400, Christopher Hicks wrote: > Sorry for beating the dead horse a little more, but here goes... > > On Tue, 5 Aug 2003, Nicholas Clark wrote: > > But *do not* send out an "all's well" message, which will get filtered > > with the spam to /dev/null, because crying wolf like this will cause > > people to miss subsequent real, serious, messages. > > Sheesh. It's not crying wolf. It's saying "here's the status of your > stuff". When you get a monthly statement from your bank they're not > crying wolf. They're keeping you up to date. My accountant is hopefully > more on top of things than the bank so the only purpose in the statement > is to make sure the bank hasn't screwed up, but I don't accuse them of > crying wolf because they want to keep me appraised of things.
Have you heard? Generalissimo Francisco Franco is still dead. See http://www.wikipedia.org/wiki/Generalissimo_Francisco_Franco_is_still_dead for the latest news. Periodic e-mail messages don't do anything to stop bit rot on the CPAN. They just let us shave it off and sweep it under the carpet. That seems counterproductive, especially since the correlation between having an inbox and maintaining a distribution isn't proven. On the other hand, the CPAN testers help authors maintain good code, and they help consumers find good modules. They do the former by sending failure notices to authors when new uploads are broken. They do the latter by providing reports that consumers can use to evaluate distributions. They don't prevent bit rot in stuff already on the CPAN, however. I think they could, but it would require them to retest dependent modules whenever a dependency passed its tests. The failure notices would alert authors that their distributions are succumbing to bit rot. Theoretically, they would care about their code and update it in relatively short order. Likewise it might be useful to send an automated message to the dependency's author: Your upload, FooBar-12.99, has caused new test failures in the following distribution(s): .... Deadbeats wouldn't bother, and they'd receive correspondingly more notifications. "Goofus doesn't update his CPAN distributions and gets a lot of e-mail about his broken code. Gallant keeps his code up to date, and lives in relatively unaccosted bliss." Which brings us back to e-mail validation. Since the testers are sending out all these notices, perhaps they could monitor and report on bounces. Other systems, like PAUSE, could use the information to prompt authors to fix their e-mail addresses when they log in. -- Rocco Caputo - [EMAIL PROTECTED] - http://poe.perl.org/
