Re: eval-when, eval-when-compile

SJT> I will say, as an aside, that the last few months of hacking SJT> on the package build process have been a massive step SJT> backward. :-( mb> Paging Steve Youngs... mb> I have certain release criteria for 21.2 beta, like successful mb> C++ compiles and clean `make check's. I would hope package mb> releases would have similar release critieria. Web page! Web page! Write them up, Martin. I don't think this is a Steve Youngs problem, unless he wants to claim it. We don't have a clear statement of what the package build system is supposed to do besides providing individual package tarballs and SUMOs. AFAICT, Steve is doing a pretty good job of the well-defined task of rolling out tarballs; the recent spate of re-re-releases has been more related to changes in the internal organization of certain packages and to the fact that Steve can't how know all those things affect package builds without a certain amount of cut-and-try. This is harder to automate at this stage than the criteria you set for the core. (IMO) The problems I mentioned are different, and are organizational. The missing statement of purpose. We don't have a review procedure for changes to the package infrastructure. I don't know whether the package system should be treated like the stable branch, with a single gatekeeper, but we should consider that (if we can find somebody who is willing to own the Makefiles, package-compile.el, etc). I personally would like to see a lot more control on the kinds of hacks that karlheg and Adrian have inserted into the build process recently. We need a Mule policy; the lookup snafu shouldn't have happened, and it shouldn't have persisted. (This one is really my bad; as Hrvoje would have it, it's Mule vs. XEmacs and I should think as much about protecting no-mule xemacsen from Mule stuff as about extending and improving Mule itself. Also packages have always been another Steve's bailiwick ;-) -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."