Date: Wed, 26 Jul 1995 23:29:10 +0930 (CST) From: Michael Smith <msmith@atrad.adelaide.edu.au> To: asami@cs.berkeley.edu (Satoshi Asami) Cc: msmith@atrad.adelaide.edu.au, jkh@time.cdrom.com, rgrimes@gndrsh.aac.dev.com, current@FREEBSD.ORG Subject: Re: Knobs in /etc/sysconfig Message-ID: <199507261359.XAA05059@genesis.atrad.adelaide.edu.au> In-Reply-To: <199507261206.FAA10240@silvia.HIP.Berkeley.EDU> from "Satoshi Asami" at Jul 26, 95 05:06:47 am
next in thread | previous in thread | raw e-mail | index | archive | help
Satoshi Asami stands accused of saying: > I'm not sure why you say so. My guess is that for most of the ports, > it doesn't matter in which order it's started (relative to the other > ports daemons). Only a few of them would actually depend on another > port, and we'll have to be sure we assign numbers to them in correct > order. > > And the numbers don't have to be unique. In fact, I think most ports > can have the same sequence number, the canonical "50" or something > like that. Until you find yourself in the situation where A & B are independant of one another had have the same number, but the new port C has to fit between them. The solution is close, but not adequate. > * If you can handle keeping a list of 'what goes before/what goes after' > * for each port, then a sequentially arranged list in a control file > * could be realistically managed, with entries inserted and deleted as > * packages come and go. > > You make it sound so easy. How are you going to go about that? > Please write a sed script for me. :) I'm not a sed guru, so I'm not going to humiliate myself 8) It'd be relatively simple to implement with a linked list in any structured language. > Also, unless we add another program to the system that handles this, > we'll have to let the porters write sed scripts. Even if we have a > template, this is bad, as someone can make a mistake and totally screw > up the system. It would fit into the dependency concept that we already have; you list for a given port anything that must preceed or follow it, nothing else. Then, when something gets added to the file, you skip lines in the file until you pass everything that you must follow, meet something you must preceed, or hit the end of the file, and insert it there. (Then you scan the rest of the file to make sure that its consistent with your view of the world. There's scope here for the A,B,C situation described above to cause grief, but if the preceed/follow lists for all the ports are available the reshuffle can either be resolved or you have an inconsistency in the ordering.) > Making automation easier is the whole point of the ports/packages. :) > Also, I don't understand why this would increase the pain of work by > hand. If the files are plaintext (as would probably be the case), you open yourself up to a much greater risk of inadvertent damage as you increase their structuredness (8)). > Satoshi The question remains; how carried away do we want to get about this? I'm generally fond of doing things "properly", but that's as much a matter of perception as anything. How is this likely to relate to the assumedly imminent restructuring of the package management tools? -- ]] Mike Smith, Software Engineer msmith@atrad.adelaide.edu.au [[ ]] Genesis Software genesis@atrad.adelaide.edu.au [[ ]] High-speed data acquisition and [[ ]] realtime instrument control (ph/fax) +61-8-267-3039 [[ ]] My car has "demand start" - Terry Lambert [[
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199507261359.XAA05059>