On Sat, Mar 22, 2003 at 12:13:05PM -0600, Manoj Srivastava wrote: > Were there any normative changes made whan this restructuring was done?
A few non-controversial things, I think. I added all the fields (like Closes, Changed-By and possibly others) to the Policy and described those that weren't in the Packaging Manual. I don't believe that this needs anything more than proofreading to make sure I got the list right, and a few seconds to confirm, because we already had most of not all of those fields mentioned in the Policy, they simply weren't nicely listed. The fields that weren't mentioned at all (I think only Changed-By) are listed as optional, so they can't possibly cause any harm. I also added a sentence in the section about the non-standard changelog formats from the packaging manual that said that the new formats have to follow the same ABI that the dpkg-* tools use. This also goes without saying, if they didn't follow the ABI, the package couldn't be built... I think Julian merely omitted that particular sentence by mistake with the initial merge. And lastly, I found a bug in one of the old commits: 1.64, which added the parts of the build-arch and build-indep stuff into the Packaging Manual appendices instead of putting everything into the Policy section. Technically this would be a three-paragraph change in the Policy but to go through the same process again would be rather anal-retentive. :) On that note, by proofreading the debian/rules sections I determined that the last restructuring caused bug #88029. Not that there was anything wrong with the method used[1], it merely made me lament the fact I didn't proofread that one back then, and made me be more careful about the may-should-must phrasing now. Which reminds me, I'll go make a diff -wu (ignoring changes in whitespace) and post that, it will likely be more readable. > And which version of policy was the restructured document forked from (a > CVS tag would be nice, so we may track the changes made in the current > policy doc that have to be moced to the new layout). Oh, the very latest CVS version. I noticed your changes a few hours/days ago and merged them in before posting. (I'll do so with any other changes anyone commits, presuming they don't rewrite the whole thing from scratch. :) [1] I saw clear pattern in how the various "is" and "should usually" terms were converted into "should" and "must". The conversions were consistent, however, IWJ wasn't necessarily consistent in the original text because the original text was never scrutinized in this manner, to my knowledge. This, combined with the fact the rules file interface was obviously made in a way a that doesn't really rely on make's features, has convinced me even further that the "must" rule (which I objected to in #88029) is improper. :) -- 2. That which causes joy or happiness.