Date: Sun, 10 Sep 2000 18:24:32 -0500 From: Will Andrews <will@physics.purdue.edu> To: Satoshi - Ports Wraith - Asami <asami@FreeBSD.ORG> Cc: Jeremy Lea <reg@FreeBSD.ORG>, Bill Fumerola <billf@chimesnet.com>, Will Andrews <will@physics.purdue.edu>, FreeBSD Ports <ports@FreeBSD.ORG> Subject: Re: Ports Options Paper Message-ID: <20000910182432.C1309@radon.gryphonsoft.com> In-Reply-To: <vqc1yyrub1y.fsf@silvia.hip.berkeley.edu>; from asami@FreeBSD.ORG on Sun, Sep 10, 2000 at 03:39:05PM -0700 References: <20000903052226.E1205@radon.gryphonsoft.com> <200009082243.e88Mh9V05579@silvia.hip.berkeley.edu> <20000910175639.F47559@jade.chc-chimes.com> <20000910152633.E39816@shale.csir.co.za> <vqc1yyrub1y.fsf@silvia.hip.berkeley.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Sep 10, 2000 at 03:39:05PM -0700, Satoshi - Ports Wraith - Asami wrote: > * How about looping through all the combinations of WITH_* (or how ever > * the options are maintained) and then only building the packages for > * which the package name changes (via PKGNAMESUFFIX), which have the least > * WITH_* defined. > > People, we have done that before. Is it too much to ask that you go > read the cvs logs of bsd.port.mk and the mailing list around October > 1998 before suggesting something that has already been tried (and > failed) in the past? :< I only briefly reviewed the diffs between bsd.port.mk revs 1.293 & 1.294. I didn't notice anything in particular that could have broken things, but perhaps you could clarify. All I see that we'll have to do is set up a variable that specifies the options the package cluster will build (and can wrap this depending on the arch), where it will then do a package build for that particular option set. I think the model of looping like it did back then is not the right way. The right way would simply be to obtain the list of PACKAGE_BUILDING option sets in a string and having the package building script loop between them. And if you modify WRKDIR to differentiate between different PKGNAMEs you'll be in the clear, as long as dependencies are satisfied. ``make describe'' can be modified to output 2 fields: available option sets, and package-building option sets. In other words, I don't like the idea of looping being built into bsd.port.mk. > I know having some extra directories is an eyesore to some people, but > that's a relatively minor problem, and we have much bigger fish to fry. It's not just an eyesore, it results in additional unneeded inodes. It's just too inefficient and if we stick with your philosophy, things will be so bad by the time we reach 7,500 or 10,000 ports that we're not going to be able to reconsider. I want to avoid a catastrophe and make it as efficient as possible. I don't really care about an extra 15 or 20 lines of code or changes to the package build system, because they can be done a lot faster than if we had to reconsider this whole issue all over again in two or three years. Hell, at the rate of about 100 to 150 new ports a month, we're going to pass 10,000 in less than three and a half years. And that's ignoring the fact that ports have grown at more of an exponential rate than a linear one. -- Will Andrews <will@physics.purdue.edu> <will@FreeBSD.org> GCS/E/S @d- s+:+ a--- C++ UB++++$ P+ L- E--- W+ N-- !o ?K w--- O- M+ V- PS+ PE++ Y+ PGP+>+++ t++ 5 X+ R+ tv+ b++ DI+++ D+ G++ e>++++ h! r- y? To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ports" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000910182432.C1309>