Slashdot Mirror
Linux Turns 25, Is Bigger and More Professional Than Ever (arstechnica.com)
The Linux operating system kernel is 25 years old this month, ArsTechnica writes. It was August 25, 1991 when Linus Torvalds posted his famous message announcing the project, claiming that Linux was "just a hobby, won't be big and professional like gnu." From the article: But now, Linux is far bigger and more professional than Torvalds could have imagined. Linux powers huge portions of the Internet's infrastructure, corporate data centers, websites, stock exchanges, the world's most widely used smartphone operating system, and nearly all of the world's fastest supercomputers. The successes easily outweigh Linux's failure to unseat Microsoft and Apple on PCs, but Linux has still managed to get on tens of millions of desktops and laptops and Linux software even runs on Windows.Do you use any Linux-based operating system? Share your experience with it. What changes would you want to see in it in the next five years?
You are confusing contributing with leading the project.
I remember when Linus posted it. I downloaded it and played with it a bit.
When Slackware 0.99a came out I gave it another try. It was not long before I was converting my Minix boxes at the house over to Linux.
In 1995 I switched from Windows 3.11 to Slackware and never looked back. To this day I run linux on all my systems at home save a small laptop that runs Windows XP though it is just to manage the spectrophotometer which does not have a linux driver.
Linux has come a long way and I am always amazed at how much of the world runs linux from Cell Phones, to routers, to supercomputers.
Ardour is great, and so is Reaper. The existence of a solid DAW on Linux isn't the issue at this point.
First, one of the major issues is inertia - Logic Pro, Ableton, ProTools, Cubase, Sonar, and FL Studio are all respected names in the field, with lots of users, forums, and ecosystems around them. Audio engineering is very susceptible to a herd mentality, because anyone who uses something different will be told to join the herd, rather than getting actual support.
Next, audio engineering is much more hardware dependent than most CS/IT disciplines. For us, 'input' basically consists of keyboards and NICs, which are interchangeable. Pro audio involves audio interfaces from Tascam, Presonus, M-Audio, and FocusRite, with MIDI controllers ranging from Korg/Yamaha keyboards to guitar pedals and drum pads. We'll circle back to the interface problems in a moment, but the MIDI controllers are largely USB now, meaning there are abstraction layers that may require specialized drivers, mapping software, and plug-ins.
Back to the audio interface question, amongst the major things we have here is that Jack/Alsa are fine for desktops with Realtek chipsets, but when you're dealing with thousand dollar interfaces that can record sixteen channels of audio in real-time with 1ms latency, Jack and Alsa just don't cut it. OSX has CoreAudio and Windows has ASIO, both of which are industry standards that work with those interfaces. Linux would need something similar to it, but even if such a thing were to come into existence, support by the hardware OEMs is certainly not coming into place overnight. Meanwhile, those OEMs need to sell gear, which means that CoreAudio and ASIO handle over 99% of the market, and no one seems to be chomping at the bit to write yet another audio system for Linux to even provide a viable target. Reaper and Ardour could well start on that, but now you have DAW devs stuck writing middleware that already exists on Windows and OSX.
I look forward to it happening, but it's a pipe dream right now. Hardware OEMs are targeting ASIO and CoreAudio, plug-in writers are targeting Ableton, Protools, and VST hosts, industry standard DAWs are targeting Windows and OSX, and a soup-to-nuts Linux ecosystem would require cooperation from everyone at the same time for a market segment that's super picky at best.
Wake me up when even Windows follows that paradigm. In fact, Microsoft is, at least in the enterprise, moving explicitly away from the GUI, and pushing Powershell for many tasks. But really, it's always been that way. GUI configuration tools in Windows have always presented only a portion of the configuration options, and many settings have had to be adjusted via the Registry. Even with GPOs, many settings can only be accessed via the Registry.
Like any system, whether it be Windows, OSX or Linux, everything works great out of the box... until it doesn't, and at that point the user is forced to go to some pretty daunting places. I've had enough fun trying to install drivers in Windows, or trying to solve problems on everything from screwed up profiles to getting the damned thing to time sync properly to know that Windows "ease of use" is more a marketing slogan than reality.
The world's burning. Moped Jesus spotted on I50. Details at 11.
The other problem is resistance in the Linux community to complex tools - because the problems are complex to solve. Even if you apply the "do one thing and do it well", it ends up as a complex tool (see SystemD). And no, sysvinit scripts are not the solution (question - why does /sbin/init provide a perfectly usable daemon manager that no one uses? I mean, it will monitor daemons, if they die, it will restart them. If they die too quickly, it will pause restarting to let the admin have CPU time to fix the problem).
System initialization isn't easy - Apple has tried many different forms of system initialization daemons until settling on launchd (they started with sysvinit at first, then migrated to SystemStarter and a couple of others). And the BSDs have tried to port launchd over as well.
Then there are other use cases - networking for example. NetworkManager is a solution to a problem users have - they may connect to different networks with different network settings. Because without it, handling the simple case of a user going from home wifi to public wifi is much harder. At least to Linux's credit, when it detects public wifi, it can auto-start a VPN client, or even prevent unencrypted traffic in the narrow window between connecting to public wifi and before the VPN starts up. Or even something as minor as going from static IP to DHCP.
Then there's PulseAudio, a framework made necessary because users are complex. Such as being able to switch audio devices while the program has the audio device open. E.g., VoIP - user might be having it on the main audio device waiting for it to ring. The moment it does, users plug in a USB headset (new audio card), and have the call audio automatically routed to the headset without the controlling application (VoIP program) having to do a thing. Or a user switches from onboard audio to a Bluetooth headphone and being able to do it transparent to the player application.
Of course, there's a Linux that does all this transparently to the user - we call it Android. And all this stuff is complex because it has to be - there's no simple way to have a system do these tasks.
We are still seeing this shit after MS Powershell came out?
Look up "grep", "sed" and "awk" and you'll see why some people dealing with CSV files or similar are happy that there is a command prompt instead of having to wait for someone to write a special program for them.