But what was their reason to reject it?
Posted by damog at Mon Aug 8 02:48:02 2005

The .reason file says: "As written in the mails: I do not think it is good to split the 1.9 and 1.6/1.8 packages".

I think Ukai-san's characterisation is pretty unfair.
Posted by Colin Watson at Mon Aug 8 03:08:56 2005

Colin, I wonder if it is.  I know that certain Joerg's responses as an ftp-master have been quite a bit more harsh and have been enforcing rules that are not based anywhere in policy.

This came up when talking about some "unfortunate" features in cdbs.  It's bad enough that the buildds don't follow policy, but it's a problem if the ftp-masters start to enforce their own version of best practices that aren't actually documented anywhere.
Posted by Jeff Bailey at Mon Aug 8 05:40:20 2005

Jeff: Well, rejects based on a particular aspect of the package (as opposed to the fundamental nature of the package) are rarely permanent. You can discuss it with the ftpmaster in question, fix it if necessary, and get on with life. It's on that basis that I thought the tone of Ukai-san's entry was unfair: it's not permanent hindrance, and the issue could for example be avoided by dropping support for some older versions of things that should be being phased out (after all, we don't want to keep old versions of everything around forever), or by merging some of the packages together.

In the particular case of gonzui, the applicable reject reason was the one I cited above, and (without having seen the mails in question) seems like it could be resolved by either merging libdb4.3-ruby1.9's binary packages or providing good technical reasons why they can't be merged.

I agree of course that best practices should be documented better, but that seems kind of orthogonal to this case, which is just good old-fashioned "too many binary packages". :-) It's rather different from the "this is known not to work, so don't do it" rejects I think you're referring to.
Posted by Colin Watson at Mon Aug 8 11:43:24 2005