| |||||||||||||
Where have all the reviewers gone?Where have all the reviewers gone?Posted Sep 14, 2006 7:03 UTC (Thu) by dark (subscriber, #8483)Parent article: Where have all the reviewers gone?
I think you should distinguish between review before a patch is accepted, and review after code has been published and is in the hands of users. Only the latter is what I would call "peer review" -- since it's independent -- and it's what the often-proclaimed advantage refers to: our code is public. And that's the published code, not proposed code on a mailing list.
Otherwise, proprietary software could have the same kind of "peer review", and many of them do have proposed code reviewed by fellow developers before it goes in. (Well, ok. Some of them. :-)
(Log in to post comments)
Where have all the reviewers gone? Posted Sep 14, 2006 7:54 UTC (Thu) by tjasper (subscriber, #4310) [Link] I think this is a valid point. Whilst the quality of code review before the code gets accepted is raised by good reviews, if the code gets out and is still buggy, then users with enough of an itch will patch it so it works for them.
So maybe the balance has to lie somewhere between perfect code which has been reviewed to death so that there are no bugs left!!! and obviously buggy code which would bite users too much. I think the community needs to recognise that stifling new code because no-one can review it deprives the community of it's best weapon - all the users who find bugs and help fix them. Sure, overly buggy code scares users away, but there is a balance - which is probably different for every project......
Hmmmm.....
Where have all the reviewers gone? Posted Sep 14, 2006 8:04 UTC (Thu) by tjasper (subscriber, #4310) [Link] Consider further that Linus himself has stated that some of his very early code was ugly IIRC! So it does it have to be perfect every time? It was Linus' release early, release often approach which excited people as they saw new stuff appearing quickly.
The new kernel development model does the same, and it may pay for bigger projects to look for ways to get new code out there being used by people who find bugs through use rather than review before the code is released.
Also, the Coverity checker seems to be doing a good job of improving code quality without the backlash from the authors (hard to insult a program!! :o) ). Long may that continue.
Where have all the reviewers gone? Posted Sep 14, 2006 22:39 UTC (Thu) by dlang (✭ supporter ✭, #313) [Link] this is a very important point
reviewing patches to see if they are acceptable requires not just someone who can understand code, but also somone who's judgement is trusted by the project. Earning that trust is hard, and people who do are frequently authors of code rather then reviewers of it.
on the other hand it takes no credentials to review code once it's been published. anyone can do it, and if they spot a problem they can report it (with or without a patch, it's perfectly fine to say 'hey, this looks wrong' even without a patch to fix it, these are bug reports, not feature requests)
this doesn't mean that the number of reviewers of both types can't be increased, we need more of both types, but the problem isn't the pending disaster this article implies.
|
|||||||||||||