On Mon, 23 May 2005, Dirk Eddelbuettel wrote: > On 23 May 2005 at 19:15, Don Armstrong wrote: > > What I'm talking about (and why I'm continuing this discussion) is the > > optimal resolution of this issue, post sarge release, by fixing: > > > > 1) r-base-core to provide the correct headers/fix whatever > > headers are supplied > > Now, as I mentioned earlier, as of R 2.1.0, the API has seen some > cleanup; R-2.2.0 may better that. So we'll know more in October (or > from the SVN checkouts / r-devel tarballs).
Yes, which makes the closure of the bug rather strange, since this API cleanup has the potential of addressing this bug. You're saying that the API as distributed now is correct, because Startup.h isn't in the API. That's fine, and I've no argument about that. I'm saying that in the future, Startup.h should be included in the API so rpy and other modules that would otherwise need to include their own copy of the headers can be built properly. These are not diametrically apposed statements. > > That basically precludes any further discussion, and will likely > > Exactly. :) That's frankly rather rude. What makes is even more rude is that I'm just trying to help make a package that I don't particularly care about (rpy) better.[1] > There is always the web archive, though. Google doesn't forget ... The BTS is not crawled by google. > | > I do not think that changing the set of distributed header files is > | > a good idea. > | > | Can you explain why? > > R is *beautifully* cross-distro, cross-platform, cross-anything. I > do not think that making out R different adds value. Quite the > contrary. > > (I am already on the edge by having "r-base" split off r-base-core > and all the other packages. As per some R Core members, a 50mb sumo > package would be preferred. I disagree there, but I don;t want to go > much further than that.) Just because other distributions and platforms are not capable of dealing with packages in a sane manner does not mean that saddling users of R with pointlessly gigantic packages is a good idea. It may be best in the eyes of upstream, because then they don't have to worry about modularization, but it's not necessarily the best thing for the users of Debian packages. Just compare the R situtation (a monolithic upstream package, with an upstream who isn't enthused about modularization) to X's situtation (a monolithic upstream package, with an upstream who until recently wasn't interested in modularization.) What would you rather have as a user? Don Armstrong 1: As I said before, the only reason I've even entered into this discussion is because of the RC bug on rpy, and the fact that it was holding up the release. -- Miracles had become relative common-places since the advent of entheogens; it now took very unusual circumstances to attract public attention to sightings of supernatural entities. The latest miracle had raised the ante on the supernatural: the Virgin Mary had manifested herself to two children, a dog, and a Public Telepresence Point. -- Bruce Sterling, _Holy Fire_ p228 http://www.donarmstrong.com http://rzlab.ucr.edu -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
