Date: Fri, 5 Dec 2003 07:29:17 +0100 From: Clement Laforet <sheepkiller@cultdeadsheep.org> To: Mark Linimon <linimon@lonesome.com> Cc: ports@FreeBSD.org Subject: Re: [DRAFT] ports contributor's guide Message-ID: <20031205072917.3c3573c8.sheepkiller@cultdeadsheep.org> In-Reply-To: <3FD013E7.7080302@lonesome.com> References: <20031205025342.04faf48b.sheepkiller@cultdeadsheep.org> <3FD013E7.7080302@lonesome.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 04 Dec 2003 23:13:11 -0600 Mark Linimon <linimon@lonesome.com> wrote: Hi Mark ! > My main puzzlement over this (after agreeing with most of it) is that > it's kind of a surprise to me to see the emphasis on not using the PR > database for the first bug report. Now, clearly, sending in a PR > without Cc:ing the maintainer is kind of rude (although everybody does > it -- I've done this as well), but it does seem to me that since GNATS > is the "one true place" to go to look for bug reports, that we ought > to use it as such. > > In particular, if we only send bug reports to the maintainers first, > how is someone who finds a bug to know whether it has already > been sent to the maintainer or not? And if it has, how can they get > a copy of the proposed patch (if any) to try it for themselves to see > how it works? You're right, when I write this doc, I was mainly focused on feature requests, non-critical update, all kind of stuff that, IHMO, make GNATS database grow unnecessarily. I think we must encourage users to contact maintainer directly before sending a PR. Yesterday (or today) we got a pretty good example of case an user can fill directly a PR: rsync security issue. rsync is a major port and a situation such this one implies a quick commit. But when I see PR overriding maintainer work, I have a strange feeling about the real role of the maintainer. A simple "Maintainer already contacted on XXXXX" at the beginning of the PR, seems to be a good deal. > So I'm not really sure I can agree with this part of the document, > but I'm open to discussion. We already had a discussion like this few weeks ago ;-) > (I should also mention that I have the ability to send email to > mainainers who are not committers, reminding them of existing > PRs in case they have forgotten, or especially, were not Cc:ed > in the first place. It got a pretty good response the last time > I ran it). As I already told you, it's a pretty good idea. > The other thing that I wanted to put in the Porter's Handbook > but got reviewed negatively at the time is a section on "ask whether > FreeBSD really needs this port". There is a cost of infrastructure > time and people's QA time to keep the ports framework workable, > and just because someone's got a Sourceforge project doesn't > necessarily mean that FreeBSD necessarily ought to have it in > there. (Basically, I want the test to be "do you think it's going to > be useful to someone else?") I agree, some ports (like simple webapps) are useless, IMHO. We should consider to remove ports more often too. I remenber I fixed dhcpconf port. This port is outdated and no more maintained. It's designed for DHCPd 2.x and buggy. Do we really need it ? > At that same time, I also got a poor reception to my idea suggesting > that if a submitter of a new port wasn't willing to be the maintainer, > then perhaps we should think twice about putting the port in as well. Unmaintained ports are required to attract new maintainers and to teach them too. Fresh unmaintained ports may motivate people. That's why I wrote submitter should look for maintainer. Good examples are pear modules ports. Most of them are officially unmaintained, but nobody wants them, what should we do in these cases, since most of them are required as dependencies. Thierry Thomas (who proposed and maintains them) would be happy if some guys take over maintainerships. > My overall statistics page actually shows a current number and > percentage for the unmaintained ports (I prefer this to "orphaned" > myself :-) ) and right now those numbers are 2572 (26.4%). > I only once did the statistic of how much more likely an unmaintained > port is to be be broken than a maintained one, but it was on the > order or 40% more likely. Most of unmaintained are outdated, we should get rid of them. OTOH, we have ports that no one wants ! courrier MTA, ntop, and many more... The main ideas of this paper were: - helping people who want to get involved in ports to find a start point - encourage new contributors to work on unmaintained ports to keep a valuable ports tree - Teach submitters that maintainers often know what they (don't) do. - improve PR quality to improve PR management It was a draft, since this a is a sensitive subject, I prefer to ask people how they like it before sending a doc PR (to change ;)). I think it's a good idea to provide a doc like this one in porter's hanbook. I promise, next version will reconsider use of PR database ;-) Thanks for this feedback ! :-) clem
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031205072917.3c3573c8.sheepkiller>