Dan Nicholson wrote: [snip]
> OK, now I see where me and you are having the disconnect. We're both > interested in tracking down the reasons for the build order and > putting them in such a way that provides the greatest robustness and > documents its exact position. However, I think that making the > changes and then testing is going to add way too many variables to the > test. Starting from a known good recipe and slowly altering would > seem to be a more prudent way, to me. > > What you're suggesting is completely changing the build (arbitrarily > making it alphabetical) and changing it until it becomes pure? Not quite. Have you looked at bug 684 yet? :) The idea is, don't change the toolchain, the toolchain is known and the backbone of the system. But as for the rest of the packages, why are they built when they are? Starting with an alphabetical order, Chris (with some small help from me) has put the packages that *need* built first way early in the build order, if it can stand (ie, no run-time/build-time dependencies) to be built where it falls alphebitcally, it stays. So it's not *quite* arbitrary - there is a method to the madness. [snip] > > Well, count me in! Doing the purity tests and figuring out all the > niggles of this baby sounds like a lot of fun. (Man, when did I > become such a nerd?) I still plan on doing a current LFS ICA test to > see where it stands, but what you and Chris have done seems like a > good jumping off point. OK. :) So you want to join my elite team of purity testers? :) (I'll send your mask and cape via FedEX.) I look forward to seeing your results. -- JH -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
