So given that, xmem can be useful as a monitoring tool. Fluffy (my main server) runs both squid and apache, and given that fluffy only has 64MB of RAM, things can get a little cramped. If I suddenly see that the whole of xmem turns blue (i.e. the swap file’s thrashing), then I know that something is odd, and I can easily find out which processes are eating up so much RAM.
I said earlier that xmem can brighten up one’s desktop. Indeed, as I use FVWM in a rather archaic fashion, it seems fitting I should like xmem. 🙂 Here’s a full screenshot showing xmem (plus other applications) in action.
↫ Thomas Adam
This is basically just an excuse to show off this awesome FVWM desktop shown off in this short little article about xmem, written by one of FVWM’s core developers. It just looks neat.
This is your daily reminder that Motif was not a glorious promised land, but rather a foul pact made by the shadowy cabal of OSF/1 members not to backstab each other by hiring a UI design team, so they could focus on competing with each other through hardware improvements and continue putting off the very important work of actually supporting application development. CDE set the Unix world back by more than a decade!
Greetings! I am certainly too young to understand this but I am genuinely interested in it. care to elaborate what you mean please?
I don’t think I’m much older than you; this is something I’ve learned mostly from reading history. And… now that I look at it again, it seems I had a couple errors in my first post. Let me correct them…
In the late 1980s, before POSIX was a thing, being a Unix-compatible operating system didn’t mean much; every vendor had completed the OS in different ways, so applications weren’t portable between the many flavors that existed at the time. The market leader was Sun, so seven of the other major market players joined forces and worked together on a set of standards under the moniker of the Open Software Foundation—which, it should be stressed, was not “Open” in any modern sense. The main fruit of their labor was an operating system called OSF/1, which, being the product of a committee of vultures, had many questionable decisions in it, like using the (deeply flawed and slow) Mach version 3 as their standard kernel. OSF also originated the Motif toolkit, since XWindows had lacked a good standard toolkit up until that point. (The ‘portable Unix’ philosophy doesn’t work very well when you need to recompile against proprietary static libraries that your customers may not have access to.) In response, AT&T, Sun, and a few less memorable companies formed a competing cabal called Unix International. OSF and UI were not friends.
Time passed, and Microsoft poached Dave Cutler from DEC, and Dave Cutler worked his magic and concocted Windows NT. (By which I mean NT 3.1.) This set off major alarm bells at both OSF and UI, because the Windows brand had such strong consumer recognition by 1993, the two bitter rivals shook hands and merged into a sort of grand alliance of Unix vendors called COSE, which was effectively OSF on steroids. Every other major Unix vendor that had hitherto avoided involvement also threw in their lot—SCO, HP, IBM, and USL also joined. It was under the banner of COSE that CDE was developed. (There is more history after this, but it’s not as interesting. Soon they went back to the OSF name.)
The thing is: both Motif and CDE remained closed-source, proprietary software for decades, just made available from multiple companies. Because of this it was theoretically possible for application developers to target all those OSes simultaneously (even VMS has a CDE variant, and it’s not even Unix!), but you couldn’t actually go and slap it onto a new Unix unless you were able and willing to pay the COSE/OSF (or, later, The Open Group.)
So essentially that ecosystem was an oligopoly, and like all cornered markets, they had little to no incentive to develop it further. As Adurbe points out, many of the “real” applications that sold Unix systems in this time period were actually console-based or server products, so you can imagine there was probably a perception among OSF members that a GUI was just something they were obligated to ship by the market, not something they really wanted to do.
Conversely, GUI application developers just avoided the whole mess; unless they were making an exceptional product that required a high-performance workstation, they’d already been scared off Unix as a platform. Probably the majority of commercial software that actually uses Motif fits this description—scientific software that was bundled with a high-performance workstation. I remember hearing a horror story about a decade ago from an American who had to use a piece of spectroscopy software that only ran on IRIX—in German.
Application developers wouldn’t bother with Unix support in large numbers until Java, GTK, and Qt showed up with promises of truly being able to port your code anywhere, without any of the encumbrances imposed upon them. GUI application developers on other platforms—Windows, Apple—didn’t have to deal with this problem because they had way larger markets and no internal fragmentation. Out of all the Unix vendors, the only one attracting application developers to their software platform (rather than their hardware) was NeXT.
The shenanigans of the OSF were a big part of what motivated the growth of the GNU Project in this era. Had Motif actually been open-sourced from the start, most likely projects like KDE and GNOME wouldn’t have gained a foothold.
Much appreciated, I tip my hat! Cheers!
I still think the death of CDE as a concept is what set us back.
You can argue the pros and cons of the interface itself, but from an enterprise standpoint it was excellent. No matter the vendor, the interface was the same, minimising the learning curve when switching between them. All gui apps were developed to look and behave consistently in CDE.
These days, Linux desktops are massively more advanced ofc. But multiple competing desktops splits resources. Meaning apps behave differently and look different between Gnome/KDE/others. How much effort is wasted and/duplicated to facilitate the choice? Mint even created Xapps to bridge some of the gap but found it to expensive to maintain.
Also bear in mind the context that the GUI was (and kinda still is) and unnecessary novelty for a UNIX systems controlled almost exclusively via CLI.
You’re not wrong, but see my reply to Andreas; CDE’s doom was sealed from its inception due to the same greed that brought it into existence.
Is anything other than Gnome used on enterprise Linux desktops?
The last time I saw anything else in the enterprise was KDE 3,x nearly 15 years ago, and I only remember it because it was an outlier even back then and also because of the way the user slammed their mouse every time they clicked something.
I know for a long time, Advance Auto Parts here in the US used Red Hat Linux throughout the company both in the corporate side and at the actual store point of sale. Back then they used KDE 3.x for the desktop. I don’t think they ever moved away from Linux altogether and their POS looks remarkably similar to how it did 15 years ago when they were definitely still using it. Apparently they are still working with Red Hat at least as of last year:
https://siliconangle.com/2023/05/30/advance-auto-parts-taps-red-hat-for-it-infrastructure-modernization-rhsummit/
I was hopeful that my pharmacy software vendor would switch our workstations from GNOME 2 to XFCE when GNOME 3 came out, but instead we were ‘switched’ to MATE on our Fedora based workstations. I am surprised how many Enterprise Linux workstations are still GNOME based considering how many people I know that prefer other environments. Of course, I would rather use MATE or GNOME whatever the number is now than Windows 11 or Mac OS 11 so no real complaints from me.