Re: CVS Account Request

At 00:24 16/11/2000, Rasmus Lerdorf wrote:

Part of the reason the PHP project has been successful is precisely because the barrier of entry has always been nice and low. This applies both to the language itself and also to the development of the language. I can count on one hand the number of times we have had bad commits over the last 5 years.

Your hands must have changed a lot since I last saw you :) There were dozens.

By empowering people to contribute, you encourage contributions. I would rather have a badly written extension than no extension. Badly written extensions spur other people to pitch in and improve it and the feedback provided to the author helps the author improve.

It's a double edged sword. In general I agree with you, but since end users have no way of knowing what is stable and what isn't, what's secure and what isn't, there's a problem with this as well. Admittedly, I don't have a good solution for this one.

As far as a terrorist cvs committer goes. I think you are being overly paranoid. And I would challenge the statement that an unknown can sneak a harmful commit into PHP that is not caught by anybody. I read all commit messages. And I bet I am not the only one. Sure, I skip over some commit messages from known people in areas I know little about. But someone I don't recognize comitting any sort of code automatically triggers a full review for me.

So, did you find the buffer overflow bug in Onn's fribidi extension? You know what they say, you're only paranoid until things actually happen. I don't think I'm paranoid at all. If I was paranoid, I would have actively worked towards this much more intensively. I think it's not a good idea to keep things the way they are, regardless of whether or not they worked in the past years. PHP today is a couple of orders of magnitude bigger than what it was 3 years ago, and our CVS access routines remained the same.

Let's not raise the barrier of entry without cause. Show me viable cause and we can talk about specifics.

I think I made my point when I said we don't need to wait to be hacked before we take measures to reduce the chances of this happening. Things aren't too secure in this world as it is, I don't see a reason to make it easier to screw us up.

I do agree that Pear needs to be split away at some point when it is mature enough to split. And applying some ACL's to separate doc, web and src code probably makes sense as well. But let's not take this further and start changing something that has worked very well for many years.

Well, I'm quite happy with applying ACL's, assuming it does what I think it does. I don't have any sentiments for the split CVS roots solution, something that would give an equivalent state would be fine as well. Zeev -- Zeev Suraski <zeev@zend.com> CTO, Zend Technologies Ltd. http://www.zend.com/