Demolishing FreeBSD

Status
Not open for further replies.
H

hruodr

Everything that a normal user does not use, must be deleted from base. And normal users are all Desktop users,
hence, in the name of Desktop users I tell you, FreeBSD base must be filled with Desktop (in)utilities.

I once installed 386BSD. There was a (for the time) full functional DNS server. That was deleted from FreeBSD. Soon will also be deleted sendmail.

Well, why I had to recompile sendmail only for having authentication?
 
So?

When I started with the *BSDs, bind was in the base install by default. I've been using bind for ~25 years (with self-written zone files, as an authoritative server). At some point, bind was removed from base. No problem at all, say "pkg install bind", and this made the upgrade one minute longer. So I'm not complaining about bind being removed, nor am I applauding it; someone who knows what they're doing made a decision, and that decision is probably reasonable.

I stopped using sendmail about 3 or 5 years ago (I can date it, it was after we got the new sofa in the TV room, because I was sitting on the sofa for a few hours, and then switched to ssmtp). I personally think that sendmail should stay in BSD in base, because there are a lot of sites that depend on it, and because sendmail (whose history is so closely intertwined with the history or Berkeley Unix) is one of the things that defines what being "BSD" is. But of someone decides to remove it from base, I will respect that decision.

I think we're making a mountain out of a molehill, which is sort of a troll-ish move.
 
I should add, I have no problem removing sendmail. I can understand the reasoning for doing this (alternatives in ports) and I can understand the historical reasoning for it being in the base system.
Whether that decision is made autocratically by core or via the next survey, worries me not.
 
Adding to that ... first about that "desktop" strawman: There is a certain BSD with even X.Org in base. FreeBSD went the other way around: DRM drivers were in base and have been moved out, and that was a GOOD decision.

Having stuff in base has several drawbacks. If only rarely used, it takes place, time to compile, etc. It's naturally slower to upgrade. It's naturally more cumbersome to recompile with different options if needed. [...]

What base clearly SHOULD provide is a self-contained and complete "Unix" system. Yes, some form of mail delivery is needed for that, but not a full featured MTA. Simple solutions like dma weren't available a long time ago, so it was sendmail or nothing. Nowadays, sendmail is "special purpose".

What it should also provide are tools needed for maintenance and recovery. E.g. cu(1) comes to mind, as headless servers are not that uncommon, but nobody keeps a real-life serial terminal around.

And finally, it should provide SOME convenience stuff. There, a cache-only nameserver might be appropriate (but NOT a full-featured one, which is really "special purpose"). You could argue a somewhat comfortable interactive shell belongs on this list (I'm just a bit unhappy it had to be a C shell, but oh my....)

So really, stop whining and just install the ports/packages you want for your tasks.
 
ralphbsz said:
I personally think that sendmail should stay in BSD in base, because there are a lot of sites that depend on it, and because sendmail (whose history is so closely intertwined with the history or Berkeley Unix) is one of the things that defines what being "BSD" is.
Click to expand...
That sounds more like emotional than technical reasoning. Can't argue with that ;)
 
Quality standards are a lot better in base than ports. A lot of things normal users don't use should stay in base. Sendmail, however, should be replaced with DMA, because it's archaic and vulnerable. I don't understand why people are being criticized for wanting Sendmail to be replaced.

Ideally, base would fit on a single CD, but more importantly, it should be more about what's needed than MB size. It's no big deal, I can use a usb, dvd or a boot CD install to install FreeBSD. The last one I installed didn't fit on one disk. I wonder how much time is spent trimming FreeBSD to fit on one CD, than working on quality.

Also, I don't understand the title, that brings to mind a flamewar, which we don't need.
 
sidetone said:
Quality standards are a lot better in base than ports.
Click to expand...
This could be the exact opposite as well, given some "contributed" stuff that has bugs founds somewhat regularly.
sidetone said:
A lot of things normal users don't use should stay in base.
Click to expand...
I don't think a "normal user" is a good starting point here (what is "normal"?). I identified 3 rough categories of things that make sense in base. There might be more…
 
sidetone said:
I don't understand why people are being criticized for wanting Sendmail to be replaced.
Click to expand...
For my use case - it's just useful that it's in base and works out the box (including listening on port 25 which I don't think dma does?)

Laziness, inertia, a million other things to do, mean for me it will be change that means I have to do something! As I understand it, it will be for my own long-term good, but still it's a change being forced on me. Yes, I can install from ports, but it means I have to take steps that I didn't use to have to take.

I understand the reasons why it's a good idea for it to go, but can also understand why some people might not like it.
 
richardtoohey2 said:
For my use case - it's just useful that it's in base and works out the box (including listening on port 25 which I don't think dma does?)
Click to expand...
No, dma is just local delivery and, if needed, getting the mail off-site.
richardtoohey2 said:
Laziness, inertia, a million other things to do, mean for me it will be change that means I have to do something! As I understand it, it will be for my own long-term good, but still it's a change being forced on me. Yes, I can install from ports, but it means I have to take steps that I didn't use to have to take.
Click to expand...
I don't think anyone decided anything yet, but the discussion is on for sure. And sure, the long-term good here is to at least decouple the upgrade cycle (which for a complex network service could be pretty important). The minimum "price" would be a pkg install sendmail if you're fine with default build options, and of course, being able to easily build with different options (which is already possible as the port exists; you'd just have it installed twice with the base version going unused) is another "plus".

Add to that equation how many FreeBSD boxes are actually used as MTAs, and how many of *these* use sendmail for it – well.

I don't really mind it to be in base (I just build my base without it). But I really think it would be wise to remove it. Sure I don't claim to be a "representative" user, but my scenario is probably not THAT uncommon, with a bunch of FreeBSD machines (most of them virtual with bhyve or vnet-jail, but some physical as well) and only ONE of them acting as the MTA for the network. All the others are perfectly fine with dma, directing everything to that mail machine.
 
As there is so much comparison with Linux as the standard--Archlinux, one of the most popular versions, ships with a base similar to FreeBSD--even more sparse, in some ways, for example, I don't think it includes vi or wpa_supplicant. RedHat, a billion dollar Linux, offers a minimal install, also without wpa_supplicant, though it has vi. So, then it's more like comparing FreeBSD with Ubuntu. In which case, try going on ArchForums and saying, Arch must have X and must get rid of whatever. The mods will probably just immediately close the thread.
 
ralphbsz said:
I personally think that sendmail should stay in BSD in base, because there are a lot of sites that depend on it, and because sendmail (whose history is so closely intertwined with the history or Berkeley Unix) is one of the things that defines what being "BSD" is.
Click to expand...

richardtoohey2 said:
Laziness, inertia, a million other things to do, mean for me it will be change that means I have to do something! As I understand it, it will be for my own long-term good, but still it's a change being forced on me. Yes, I can install from ports, but it means I have to take steps that I didn't use to have to take.
Click to expand...

That are real big reasons to keep sendmail. The other dream with a perfect freebsd base that is better
and cleaner than every other OS, but at the end remain like every other OS.

We use Unix out of habit, and also (Free)BSD. I do think that linux do some things better, and perhaps MacOS
and Windows also. There are also better OS concepts (plan9).

Sendmail is not obsolete, no bloat, the compilation time is nothing even in a slow computer, and not
worse than other MTA. I would support the substitution of bind with nsd, that is much nicer, instead
of deleting bind, but for sendmail there is no real reason other that witch hunting.
 
So you have strong opinions and no interest or understanding of technical arguments. That's of no value whatsoever. "There's no reason"? There would be a lot of reasons, you can read some of them in this thread.
 
DutchDaemon said:
Note that this is a general user and administrator forum, where the community aims to assist those who want to install, run, or upgrade FreeBSD as-is. Discussions about what FreeBSD needs to be, or needs to add, or needs to lose, are pointless on the forums. We do not maintain the operating system here.
Click to expand...
forums.freebsd.org

Why is FreeBSD not (more) like ....

As of today, FreeBSD Forums staff will actively close down (and eventually remove) topics that serve no other purpose than to complain that "FreeBSD is not (like) Linux" (or Windows, or MacOS, or any other operating system), or that "FreeBSD does not use systemd", or that "FreeBSD has no default...
forums.freebsd.org forums.freebsd.org

Pointless discussion. Thread closed.
 
Status
Not open for further replies.
Back
Top