evan_tech -- gnome's legacy

03:16 pm, 5 Jun 05

gnome's legacy

In a recent flamewar on the linux kernel list, Alan Cox noted that some changes to the IDE layer removed a lot of functionality: "The default driver handling has been removed and half the options for obscure systems have been marked obsolete in some Gnome like purge of functionality that might scare small children."

"gnome-like purge" -- like that cascade of attention-deficit teenagers

jwz quote. Not a very good thing to be known for.

But I'm a bit hopeful from some recent discussion I've seen about Gnome 3.0. See, for example, Nat's email in response to some people discussing how to manage breaking the ABI:

[...]

Another way to look at this which I haven't even seen considered is,
"What do we need to do to create a really exciting, useful, and usable
desktop environment that people love to use?"

Certain things come immediately to mind:

        - build a baseline of useful applications that are useful,
        slick, fun, and work well together
        
        - make it possible and fun for other people to build additional
        applications that work well with our desktop and with our other
        applications
        
        - perform usability testing on what we've already built,
        identify the problems, and fix them
        
        - create compelling new and original things like what the
        Xgl/Luminocity/Cairo work might enable, like Beagle/Dashboard,
        etc.
        
Breaking ABI, changing version numbers, "rethinking everything," etc,
don't float to the top for me.

[...]

These sorts of insights are somehow simultaneously so painfully obvious, yet opaque to a large number of people (see above re: attention-deficit teenagers).

xaosenkosmos

My favorite Nat quote in some time segues nicely into what could be a good summary of his email:

Those of us who have been around software long enough feel our sphincters instinctively clench in terror when people say certain things. Things like "What is a file anyway?" or "We can just create a general system that abstracts all this away!" or "Let's just rewrite the whole thing." These are all ways of thinking that get you away from solving user problems. GNOME is primarily a project that's writing software for users, and we need to keep them in mind. Throughout this sometimes-excruciating 3.0 discussion, I haven't seen them (the users) mentioned a lot. It seems like people are more focused on being original than on building software that's really great to use.

reply to this

theorb

I'm glad that the GNOME guys are finally starting to see some backlash for removing functionality that their users use, to deal with people that might theoretically become their users at some point slightly after hell freezes over.

Mind you, I'm not against making things easier to use. I am against making things less powerful. By all means, make hard things optional. Even have an option to show more options, if it'll help.

But sometimes I see a mockup of a new UI, and think every idea on it is good for a few people, and bad for a lot of people, and most of the people it's good for will never use the software anyway, because it doesn't have a million dollars advertising it, and 5 million in legal fees.

reply to this

evan	I think you missed his point. They (and jwz) are talking about avoiding gratuitously breaking backwards compatibility. Regarding your point: you don't have to use it.
reply to this

evan	(But also to be fair, it's likely you read Alan's point as he intended it and I didn't... it's sorta two sides of the same coin.)
reply to this

theorb	Right. I care little about API and ABI compatability, and much more about compatability with the minds and instincts of your users. Example: The new file dialog in GNOME is much less bad then I thought at first. It took me quite a while to realize that you can type into it even though there is no visible text box.
reply to this

jwz	Well my point was about "rewriting it instead of actually making it work", because managing your bug list isn't sexy. But yeah, backward compatibility is also nice every now and then.
reply to this

jeffr

I think most developers have a knee-jerk reaction to want to rewrite something when it's ugly or buggy code. I find that this is more often the wrong decision than it is the right one. However, in open source, where there are no real managers, developers do what they want. In some cases, it's starting to become culturally accepted and even encouraged.

This is one of the primary reasons I stopped working on linux. Linus seems to support constant rewrites of major components. I don't see how anything can become stable this way. There are plenty of examples though. The VM, the TCP stack, firewalling code, etc.

What's worse is I've now run into people in my professional life who pitch Linux's success as validation for this model, and attempt to rewrite whole commercial products in ridiculous amounts of time. Which, of course, has all sorts of unintended consequences that are just as problematic as the original code.

Probably preaching to the choir but this is a real pet peeve of mine.

reply to this

theorb	Regarding the second half of your post: I'll try posting my rants about gnome an (the lack of) usability somewhere somewhat more relevant. (Sadly, the most relevant place I read regularly is natfriedman_rss, I doubt Nat reads the comments on LJ, and the blog itself does not have commenting enabled.
reply to this