Matthew Garrett: How-To Drive Developers From OS X To Linux

[zanny]

Steammachines are not the desktop. They are another use of linux, which, admittedly, probably uses more of the userspace than either chrome or android, but it's not a standard desktop installation.

And yes, the desktop never really happened, because everyone was too busy gutting it _all_ _the_ _time_.

KDE isn't gutting anything. Hell, even KDE4 wasn't gutting, it was just piling on a bunch of broken shit on top of something that already barely worked. But 5 years later it is pretty good, and kde5 is just an iterative improvement - making it pretty, etc, without radically changing the paradigm.

I guess we have to hope Gnome doesn't try a Gnome2->3 again, but Mate exists to keep Gnome2 alive. Which is why free software is great, because if you like some workflow more likely than not you will never have to abandon it, whereas on Windows / OSX whatever direction MS / Apple want to push the industry in they will stick down your throat, either through new releases or ending support for the version you like.

[Marc Driftmeyer]

Which is called Qt.

KDE is the example of that and it quite frankly is no where close to being in Cocoa world of robust frameworks.

[jrch2k8]

KDE is the example of that and it quite frankly is no where close to being in Cocoa world of robust frameworks.

well that is taking things a bit too far, is true frameworks mostly works all the time(IOKit and foundations always include surprises between OS X releases or even updates) but frameworks are highly counter-intuitive and global library dependencies(especially with bundles) can be a massive pain in the ass even if you let XCode manage everything and OS X in general lacks a lot in the crypto area.

In the case of Linux you got other equally serious problem which is you have too many choices and options to develop but is completely intuitive and there are extraordinary ways to handle dependencies smartly without duplicate the same library 100 times.

in the great scheme of things Qt generally does a goob job helping you dealing with this sort of problems, except for multimedia where things tend to get real hard since in Mac almost always force you to use objective-c++ or in mobile(i freaking hate ARM GPU drivers) to deal with a bazillion of driver bugs and workarounds(Qt5.3 improve a lot in this area for consumer stuff but for pro operations not so much)

[wizard69]

And in 10 more years, OS X will still be out in front for UNIX based OS platforms, more rich environment of frameworks, expanding its dominance via iOS and OS X, and Linux will still be raving about having a dozen DE that are free, but constantly breaking between upgrades.

Yep! I probably spent a good ten years trying to run Linux as a desktop OS, since well before Fedora came out, and have to say my switch to Mac OS as my primary OS was nirvana. The differences where stark and I didn't have to be on the OS upgrade breaks everything treadmill. I could and do keep my MBP current with a minimal of hassle. That usually means an Apple supplied OS upgrade every one to two years. Contrast that with Linux when a new distro upgrade implies spending months to get the platform stable again.

Face it. You've received tens of billions in developing Linux for Server Markets and consumers aren't dying to brag about the most uptime, ability to scale their LAMP set ups, etc.

And the question is why would consumers want to brag about uptime. I really don't think Linux developer have a clue about consumer needs nor a care really. That isn't bad of course as Linux is excellent in many professional applications.

Linux has never been a targeted Consumer OS Platform. To do so would require a unified set of Frameworks to develop said UI.

And a huge mind set change. However what this article seems to imply and I kinda have to agree with, is that Mac OS is an excellent developer platform. In some cases developer preferring it even when targeting Linux. This really goes to show you just how screwed up some Linux developer are when they resists the future.

It sort of reminds me of UNIX Gurus of the past complaining about HTML E-Mail and other improvements. It is like get with the program guys, the sixties passed you by a long time ago. This has been brought up because I once had a long discussion on the net with one of these Gurus about the nature of E-Mail. Apparently some of these guys never heard about evolution.

[bittwiddler]

" And in 10 more years, OS X will still be out in front for UNIX based OS platforms, more rich environment of frameworks, expanding its dominance via iOS and OS X, and Linux will still be raving about having a dozen DE that are free, but constantly breaking between upgrades."


I have to agree, Linux is essentially a niche operating system on the desktop for developers who need
extreme flexibility in customizing their development environment. Just this morning, I discovered that
my Sandy Bridge running SL 6.5 with Gnome 2 was burning 25 watts after a resume from suspend, and I had
to hibernate and then resume from that to get everything back to normal. Then, after I watched a clip
on firefox, I found out that pulseaudio wouldn't quit, and I was back up to 20 watts. Kill pulseaudio?
Ever see the movie, "Kill Bill?" You can bury it in a coffin and it figures out how to get out.

I could never use this machine for presentations on the road with behavior like that. I'm probably
going to have to get a macbook- battery glued in and all - just so that I can function in the business
world. It's too bad that there aren't enough developers out there with time on their hands to fix
stuff like this. It's not that we're a dying breed, there just weren't enough of us to begin with.

[Vim_User]

Contrast that with Linux when a new distro upgrade implies spending months to get the platform stable again.

I seriously find comments like that funny. If you have problems to get your system stable after a distro upgrade, have it ever occurred to you that you maybe use the wrong distro?

[johnc]

The obvious question is... why would somebody choose OS X when Windows is a cheaper solution and far better in just about every way anyway?

[sandy.martel]

The obvious question is... why would somebody choose OS X when Windows is a cheaper solution and far better in just about every way anyway?

"The obvious question" been answered a few times in that thread. One is Unix the other is not, one comes on hardware that seems to be generally more appreciated.

Those are 2 facts, Windows is generally cheaper is another fact, but price is not the only criteria for everyone. And Windows is "far better in just about every way anyway" is debatable (vs OS X or vs Linux).

[wizard69]

Totally agree! However the type of people who are being talked about here don't want a full blown fancy desktop with all the usability. They typically use xfce, lxde or even fluxbox and rant about how their desktop isn't too good on usability. They are so scared of losing that 1GB of ram for a reasonable desktop that they don't even know how awesome KDE is (or even Gnome3 for that matter).

Err no not even close.

[wizard69]

You have some interesting points here.

The way I understand this discussion is that we are talking about developing with not so much for. The question is why would developers prefer Mac OS/X over the alternatives.

One thing Matthew also seems to blow right by without acknowledging is that the documentation offered by Apple and MS and for them is usually magnitudes better than it is for Linux resources.

That is one thing Apple does well.

This is a key stepping stone for anyone wishing to start development for a platform. I went to four huge bookstores this week to see what was out there for Linux development. You could find isles of development books for iOS, OS X, Java and MS solutions. When it came to Linux however, the store with the largest selection of Linux books (total, not just development) was FOUR titles. Two on beginner guides for old versions of Ubuntu, one "Linux Bible", and the "OpenGL SuperBible".

The problem with Linux is how would you possibly write a book and have it be current at the time of publishing. It goes back to stability, you in many cases have to hang out on the various forums and mailing lists to have a sense of what has changed. Note the word changed here and not added.

When it comes to linux development however, you are usually dealing with outdated or poorly written documentation (rarely with even a few basic code examples) and end up praying your googlefu is strong and hopefully find the answer you are looking for in a mailing list archive or forum. While those resources are OK for an old school veteran, they are often too daunting and frustrating for a novice. Man and readme's are a shitty way of trying to encourage the novice to start development as well.

That is another interesting discussion I had years ago with one of those self proclaimed UNIX gurus. The question I ask is why are Man pages so damn sparse. They are seldom up to date with respect to implemented features and devoid of rational examples.

Then there is also the inconsistency of packaging naming and availability, file system hierarchy, multiple API/tool kits/libraries that overlap in the functionality they wish to provide (and usually none being a complete solution), sometimes rude and snotty replies from project heads towards the novice, etc.

Providing a desktop environment is important, but a properly managed documentation and support resources are more important to drawing in more developers.

True to an extent but I still see the Macs just works attitude to be a big draw. Generally you don't need to do a lot of system maintenance to keep the platform running.