One of the downsides with the open source ecosystem is that it allows the petulant child types to take their ball and run away rather than have grown up discussions to reach mutual agreement on how to solve a problem. It's very easy to get some key people and fork-off rather than have the tough conversations that may be necessary to force change. This would seem like one such occasion.
A split seems to have emerged in the Linux-router-OS community, with a breakaway group splitting from OpenWRT. OpenWRT is the chief open router firmware implementation, but it has run into headwinds of late. For example, downtime for the group earlier this year was traced back to the small organisation running a single, small …
Thursday 5th May 2016 08:46 GMT Anonymous Coward
Well, one way or another, I think the proof will be in the results.
As with any project, it usually comes down to people issues. Not everyone is a natural communicator and it takes some human skills to ensure you don't disenfranchise those.
Personally, I enjoy building good teams even though it's a bit like a fusion generator - you have to put a lot of energy in before it start to generate its own. However, once it does, stand well back.
(given the state of fusion energy I'll stick with team building :) ).
Thursday 5th May 2016 09:03 GMT Flocke Kroes
One of the upsides of free software development ...
... is that when a disagreement turns up, even if it is trivial, petty or downright stupid, a fork will lead to either two good projects that are suited to different users or to one project falling apart or getting ignored.
In computing, real learning comes from doing. Sometimes that involves doing it wrong a few different ways until you know better even if ten thousand other people have already learned from the same mistake.
Thursday 5th May 2016 17:08 GMT Anonymous Coward
Re: take the ball and run away
"the open source ecosystem ... allows the petulant child types to take their ball and run away rather than have grown up discussions to reach mutual agreement on how to solve a problem. It's very easy to get some key people and fork-off rather than have the tough conversations that may be necessary to force change."
Whereas that would never happen in the commercial world?
How many other examples are there besides Cutler taking NT from DEC to Microsoft (and look what happened to NT after that, and to DEC, and to Cutler)?
Thursday 5th May 2016 07:44 GMT Mage
I have no idea
Perhaps we will see how it works out?
I'd not automatically brand them as "petulant child types". I've using OpenWRT for over 10 years and the forums etc are not the most useful resource. It's been very quiet outside of OpenWRT. It does need to encompass a bit more than simply being a replacement for SW on a retail router. In place updating rather than create a "from scratch" install and the put back in all the old settings seems tricky.
Thursday 5th May 2016 07:50 GMT Jonathan Richards 1
You say downside, I say upside...
... but nobody calls the whole thing off.
First of all, forking a project when it has the maturity of OpenWRT is not 'very easy', I suspect. However, your analogy with a childrens' ball game is inapt. The whole point is that there isn't just one ball. The forkers are taking a clone of the ball, and starting a new game in the same field. Current and future players can decide which game to play in: hell, we can play in both, if we like. Meanwhile, there are two games going on, rather than "tough conversations" on the sidelines which cut no code.
Some forks are what you might call 'hard', with a disagreement about technological direction, e.g. the Devuan thing, but this seems to be softer, more about project management. Don't rule out a merger in the future.
Thursday 5th May 2016 12:45 GMT Anonymous Coward
Re: You say downside, I say upside...
I'm glad to see these forks. Debian has the insurmountable bureaucratic and architectural problems one should expect from a huge 2-decade-old FOSS project. OpenWRT (and DD-WRT) look pretty sketchy as OS distros go. I realize part of the problem is the profusion of more-or-less proprietary consumer router hardware, but I'm not at all surprised to hear about the political/bureaucratic issues.
If they want to do different things, they should just fork off and do different things. Competition is beautiful.
Thursday 5th May 2016 09:35 GMT PushF12
Thursday 5th May 2016 09:40 GMT Nigel 11
What we really need
What I am still hoping for, is that someone will ship a fully open low cost SoHo communications box with its entire hardware design open-specified, and with it's re-flashing system made as easy and un-brickable as possible. Oh, and while I am wishing, that they can find an ADSL2+ chip with an open-source driver, or at least an open-source-friendly manufacturer and specification.
It won't be an existing router manufacturer. They sell pretty much the same hardware at lots of different prices with the high ones justified by extra features implemented purely in the firmware.
They also have to keep the FCC happy in the USA. Heaven forbid that we write our own firmwares! But there's definitely a market opportunity here for a new start-up who fully embraces open source and doesn't write proprietary firmware at all. Sort of like a Rasberry Pi, but for networking.
Thursday 5th May 2016 10:02 GMT David Pollard
Re: What we really need
Given the numbers of XP machines that for various reasons are still running, a router that could be tightly locked down and tailored to their specific functions might well see a decent uptake. Like other open source projects, though, it might face opposition for those with vested interests.
Thursday 5th May 2016 10:26 GMT Nigel 11
Re: What we really need
If you just want to firewall a subnet, you can do it all with a PC, any Linux distro and (possibly) an extra NIC. If you already have a Linux server and a DMZ, just another NIC. And although PCs used to be power-guzzling large whirring things, this is no longer the case. You can get a fanless mini-ITX board with two GbE ports and J1900 CPU for around fifty quid. Just add RAM and a case/PSU. For a dedicated firewall boot off a USB stick, no HD/SSD necessary.
Where you start needing router hardware is handling high-bandwidth WiFi access (a USB dongle doesn't really cut it), and especially ADSL (only way I know of is a proprietary router in "bridge" mode or a Vigor ADSL PPPoE bridge/modem which costs more than many ADSL routers). Or of course, pay BT for "fiber" if its available, which does at least get you the whole internet on an Ethernet cable via BT's own bridge/modem.
Thursday 5th May 2016 12:31 GMT Anonymous Coward
Re: What we really need
I use 2x TP-LINK 722N (for redundancy) high gain usb dongles to serve dozens of peoples wifi at their office through an ODROID C2 (gigabit ethernet). Id use a Pi but the 10/100 ethernet is a bit limiting.
All housed in a bland looking white cheapo project box from RS or Maplin (cant remember where I got it) with holes drilled to mount the antennae. So it looks innocuous mounted on the wall.
Im using a variant of Debian with this: https://github.com/oblique/create_ap
All set up to automagically start up when the device boots. Ive also had an friend make a POE adapter to power it.
Never had any bandwidth problems. In fact the throughput rivals a lot of off the shelf alternatives.
This outfit does a *lot* of video stream testing.
Best thing is, it's very cheap, very secure (as ive set it up to only be configurable via TTL cable so no open ports) and very reliable.
The range is outstanding as well...I could make it better with a wifi amplifier.
In case it does fail. Theres an identical one in the cupboard that is tested once a week and can be plugged in at a moments notice.
My point being...if you want a custom WAP you dont even need openwrt or the like.
Thursday 5th May 2016 23:35 GMT Long John Brass
Thursday 5th May 2016 18:22 GMT LeoP
Not just as a router OS
I am a heavy user of OpenWRT, and I do feel the pain with the community process.
For me the most important part is not the ability to transform a router into a router, but to transform a €15,- matchbox sized something from china into something useful: Still have a physical machine somewhere because it needs to talk to some USB/Serial/whatever? Take a "tiny white box" and OpenWRT, virtualize that box and do USB/IP.
One of these days someone is gonna call it an IoT-thingy and my day will be spoilt.