I didn't know Joe's skillset and as an Apple trainer I am in the habit of playing it safe with customers - otherwise they rely on me where they don't have the skillset
In a server discussion forum?
From what I've seen here, the skill level does vary widely, but anybody who knows much about the login hook mechanism is most likely savvy enough to write or at least edit shell scripts, so it's a decent assumption that he or she can master the
defaults command or edit /etc/ttys. It's not as if setting scripts via MCX is much easier (see below). And frankly, it can be a disservice to pretend that there's an easy, GUI-based way of doing this stuff. If you need to debug your login scripts, Workgroup Manager won't help you much.
I don't issue "blanket proclaimations" or admonition anything. You're the one making blanket proclaimations that there is plenty usage of client-based loginhooks.
Could we try to not let this discussion devolve into a bunch of
tu coque responses?
To start with, you incorrectly said that there were only two ways of doing login hooks. When I corrected you, you replied with some more incorrect information, and then finished with this:
Sticking to the Apple way of doing things is often the safest way. You never know when the next system update alters things just enough to break your "workaround".
Perhaps what you said wasn't a blanket proclamation. But sniffing that alterntives to your solution are 'workarounds' (with the implication that they're not Apple-sanctioned) is ridiculous. Client-based scripts are not a workaround. If you look at the documentation (more below), Apple hardly mentions MCX preferences as a login hook mechanism at all. And the client-based 'workaround' has been an officially sanctioned way of doing things since at least back to 10.2 (don't know about 10.1, as I said before). It continues to be officially sanctioned in 10.4. So let's can it with the scare quotes and the snark, shall we?
As far as me making blanket proclamations, I said:
MCX-based preferences may do exactly what you need in your installtion [sp], and that's fine. But there are plenty of folks out there who have lots of experience using client-based login hook scripts, who have good reasons for using them.
Note the phrase "plenty of folks." Sound like a universal generalization to you?
Finally, a last nugget, after which I will abandon this back-and-forth:
Sounds more like a humble opinion to me. Have you taken a poll to verify your point?
It's not an opinion. I've spent three years interacting with people who manage installations of Mac OS X and Mac OS X Server as part of their work. I've read their posts, replied to them, and asked questions of my own. I've done this on Apple Discussions, www.bombich.com, afp548.com, and Mac OS X Labs/MacEnterprise.org. I've interacted with Apple SEs. I make my claim on the basis of a good deal of personal experience and interactions with other folks--it's not just snark. Finally, if you look at the differences between the various options, and how Apple has publicized them, you'll see that there are good reasons to make the claim:
- Before 10.4, there was no such MCX setting for login/logout hooks. As you can tell from reading the posts here, a number of people are still running 10.3.9 (or mixed 10.3/10.4) installations. Even those who are using 10.4 may not be using MCX at all, and may not want to.
- The MCX preferences settings is not well-documented. If you do a search for login hook on Apple's support site, the one
KB article that comes up doesn't even allude to it--it only discusses /etc/ttys and root's loginwindow.plist file. The only place I've found MCX prefs mentioned for this, even in the manuals, is in the User Administration guide--a grand total of about six paragraphs.
- Using MCX doesn't really simplify things that much. You have to use the
defaults command to enable MCX prefs on the client, which is about the same as...using the
defaults command to specify the client's login hook in the first place. The main difference with the MCX approach is that the scripts live on the server, not on the client--a problem that the rest of us dealt with way back in 10.2 by dispatching from local scripts to scripts on a static automount. Plus, if your dispatch script is local, you can still run scripts even when the server's not available.
THAT is why I say that there are plenty of folks who use client-based login hooks.
In the v10.2 material they used to use the specific mount tools directly. I believe there is some reason why not to use them with v10.3 and later.
That's interesting. I wonder if that means they officially deprecate the protocol-specific versions, and why (I presumed that commands like mount_afp were completely transparent wrappers for
mount, but perhaps I was mistaken). Certainly there's no reason that I can see not to use the generic
mount command, as you describe. Your allusion to problems unmounting SMB filesystems from the command line tickles a memory from some time ago, but I don't know the exact nature of the problem, or whether it got fixed.
Perhaps it's time go through and change my mount_afp calls....
David Walton