Re: Some questions on the LSB build environment
Mike Hearn <[email protected]>
| Newsgroups | gmane.linux.lsb.discuss |
|---|---|
| Message-ID | <[email protected]> |
> Heh. Bet you a nickel you won't. Might as well try. Nothing to lose :) > The packaging debates were too > painful. I did look through some of the archives for this list, going back right to the start, but didn't come across a particularly serious packaging debate that I recall. Could you give me some dates and on which lists? I'd better read up on what went before so we don't end up repeating old stuff. > If you try, please start out by coming up with a list > of concrete disadvantages to the current .lsb packaging system. I assume by .lsb you mean the minimal RPM system that requires everything to be statically linked? If so then off the top of my head: - everything needs to be statically linked (no dependancies) :o - RPMs tend to be not relocateable. Really binaries should be relocatable, there are too many advantages for them not to be. I think you can make them relocatable but all too often support for this is missing in the build system side of things. - No provision for an apt style automated installation network (i think if somebody is going to do something, it might as well go end to end and produce a comprehensive system). Obviously one step at a time, this is something that's nice to have, but not crucial. - It's hard to make portable RPMs. File paths are usually relative to root, no support for swapping binaries depending on compiler versions and so on. In general, esp given the limitations on scripts and triggers etc, it's harder than it otherwise would be to hack around non-standardised distro differences. - RPMs have somewhat limited interaction support (allowing for customisation, user feedback etc). Obviously several of those points are debatable. Maybe people think interactivity is a bad idea. I personally think if managed well, it can be useful. Maybe dependancy resolution networks are considered to be out of scope. Well, anyway, there is a discussion of the shortcomings of the present system (of which the lsb recommendations are the same but made portable) over at http://autopackage.org/faq.html - it's partly personal opinion of course so I expect I'll get flamed for it at some point, but I might as well put it forward here. > We need solid issues to chew on to overcome our lethargy > from the last few tries, and perhaps in the ensuing debate > we can infect you with our lethargy :-) Oooh goodie :) Well, I've been working on autopackage since July, and so far haven't started feeling particularly lethargic. We'll see how it goes. Anyway, I'm off to FOSDEM this weekend, so it'll be sometime next week before I can really get started...... thanks -mike