Dev.SN ♥ developers
Sections
Dev.SN
from the Boot-him?-I-just-met-him! dept.
jbernardo writes:
"Having had several issues with systemd, and really not liking the philosophy behind it, I am looking into alternatives. I really prefer something that follows the Unix philosophy of using small, focused, and independent tools, with a clear interface. Unfortunately, my favourite distro, Arch Linux, is very much pro-systemd, and a discussion of alternatives is liable to get you banned for a month from their forums. There is an effort to support openrc, but it is still in its infancy and without much support.
So, what are the alternatives, besides Gentoo? Preferably binary... I'd rather have something like arch, with quick updates, cutting edge, but I've already used a lot in the past Mandrake, RedHat, SourceMage, Debian, Kubuntu, and so on, so the package format or the package management differences don't scare me."
[ED Note: I'm imagining FreeBSD sitting in the room with the all the Linux distros he mentioned being utterly ignored like Canada in Hetalia.]
(Score: 3, Insightful) by drussell on Wednesday February 26 2014, @03:48PM
It is true that interdependencies are an issue, but they are an issue with Linux as well. You either use an old version of a library for all your installed programs or you update the library and everything that depends on it when there is a change that breaks something. I've honestly had more issues with updating things on Linux when I've used it for things than I've had on FreeBSD, but of course YMMV.
The binary updates with pkgng (which is now the default in 10-) should usually handle everything for you as long as you don't have a bunch of custom programs/builds/etc. but again, this just really depends on what you're doing, I suppose. Custom stuff relying on strange/old libraries can be compiled with the libraries statically linked in, for example, removing the reliance on something that may change for other reasons, etc.
Ports maintainers strive to maintain a given port's buildability on not only the current developers' branch (right now that would be 11-CURRENT) as well as the latest (10-STABLE) and previous (9-STABLE) 'stable' branches intended for the end-user. Often they manage to keep them running properly much, much farther back, it's just not one of the main project goals. It's not uncommon for me to painlessly build a latest version of a port with no modification on a system that is 4,5,6 or more versions obsolete. There are buildbots that constantly build these and build failures are generally fixed as soon as possible. If you've got an (perhaps 8-) or 9- box, you should still be able to just do a make install in any port directory, and it should fetch, compile, install just fine. The ports tree is one tree, there aren't separate trees for different branches/release versions but when security or other updates are applied, it IS checked to be sure it still runs on -STABLE.
My experience has been precisely the opposite. I still have internal machines here running 2.x, 4.x, 8.x, 9.x and 10.x for various purposes and reasons and I've always found it far easier to keep what I need up to date on FreeBSD than Linux. It's actually one of the main reasons I DON'T use Linux except in a few niche cases (like on one MythTV frontend box in a system for tuner card support with the other frontends and the backend running FreeBSD)... Things seem to constantly change and it's a contstant update/support nightmare! I end up manually uninstalling and re-installing things that were supposed to update themselves due to some non-obvious dependency or change somewhere breaking something else. I find the FreeBSD system MUCH easier. To me, the dependencies and such seem more clearly marked. Sometimes that means a port is marked 'broken', but then I have the option to find the broken-ness myself and fix it or work around it when I need something specific/strange/downright wierd and nutty to work the way I want.
Users that aren't so savvy shouldn't really see much difference between a 'pkg upgrade' on FreeBSD or using one of the binary update mechanisms used in the various Linux flavors. In most cases any of these systems will look at your system's version and current state, make the best decisions they can as to what should be updated and from which repository of binaries, fetch and install.
I'm sure the on Linux way is more comfortable to one familiar with that particular system just as the FreeBSD system is second-nature to me but essentially they're all doing the same thing.
(Score: 1) by drussell on Wednesday February 26 2014, @04:58PM
That was supposed to read:
Silly me, using the wrong brackets and not noticing it in preview... :)
(Score: 1, Interesting) by dbot on Wednesday February 26 2014, @10:57PM
I respectfully disagree. They are not doing the same thing at all. The key being: fixes are back-ported to a static version of a library. So imagine on your 2.1-RELEASE, there was a ports tree freeze. That is the 2.1-RELEASE port tree, forever. Now, during the supported lifetime, security issues, or major bugs occur in those ports. Critical fixes from the latest upstream source are backported down to the 2.1-RELEASE tree, for the supported lifetime of 2.1-RELASE. No major version bumps. One day, when your production schedule permits, you upgrade to 2.2-RELEASE, and now you've got all the api+config file changing updates in one fell swoop.
Right now, it's: upgrade to the latest upstream version, recursively.
I don't know if you've used Debian + apt, or CentOS and yum to use some canonical examples. The repos have stable versions of the software, and have major fixes backported, for a release (as mentioned above). That is, you run an old version, with non api/config fixes backported. Periodically, these repos are updated to more cutting edge software, so you aren't running things 2+ years old, but by and large, the software you run will be the same version until you update. So when you say more issues, how are you installing software and dependencies, on which distro?
Here are some examples of things that have been a pain over the years:
autoconf
gettext
gmake
libiconv
*perl* omfg
ruby (if you use portupdate)
php
I think by the way this is phrased, people might get the impression that they are testing things on old releases. They are testing build success, periodically, for supported releases.
Anyways, I think that this discussion illustrates pretty quickly my point. There is total failure to acknowledge that this is a problem. I believe it is, for my uses. I'm not saying They Should Fix That, because it's a volunteer project. I'm not going to do it. I'm saying this isn't even in the discussion when talking about 'fixing the ports system'.