OpenWRT Gets Forked By Some Of Its Own Developers As LEDE Project

[timofonic]

What about merging with other very interesting projects? That would be amazing!

Musl based:
- Alpine Linux.
- Sabotage.
- LightCube OS.
- Snowflakee.
- Morpheus.
- Bedrock Linux.
- DSL B.
- void linux.

Plan to switch to musl
- Aboriginal.
- Dragora.
- Openadk.

[Djhg2000]

I'm currently trying to compile OpenWRT for a test machine and I just noticed that for some odd reason the flipping make process invokes both wget and git. Probably svn as well but I just happened to catch it during a git pull and wget of some archive. Why is this not done in any of the preconfig steps or at least as a separate step before a regular make invocation?

Failed compiles are really difficult to debug when make clean wipes what looks like the entire toolchain. I also have a flaky internet connection, which is why I'm trying to compile OpenWRT to begin with (router needs a hardware upgrade and the RTL8152 driver is not in the official 15.05 x86 repo). Also the make -j1 V=s I just ran now finished without errors using the same exact .config file it failed with half an hour ago.

[starshipeleven]

the flipping make process invokes both wget and git. Probably svn as well but I just happened to catch it during a git pull and wget of some archive. Why is this not done in any of the preconfig steps or at least as a separate step before a regular make invocation?

Their make is a huge hack that pulls down source and compiles stuff to make toolchain and stuff.

It's actually cool if you don't get errors. If you get errors, its very funny (not) to get it working again as NONE will EVER give a fuck about your messages.

I also have a flaky internet connection, which is why I'm trying to compile OpenWRT to begin with (router needs a hardware upgrade and the RTL8152 driver is not in the official 15.05 x86 repo).

This isn't what you were looking for. (jedi move)
What you and most people are looking for isn't building from source, but using the imagebuilder system from trunk download page (i.e. it will build the image pulling down the trunk precompiled packages, if existing).

Much faster, but you must have a list of packages you want to integrate, which has to be gathered by hand or by looking at what is in your .config.

Dependencies work, so you don't need to specify half the system.

go here https://downloads.openwrt.org/snapshots/trunk/ select the correct architecture/device-family and look for a tar.gz package with "ImageBuilder" in the name.

Then read the wiki page on how to use it https://wiki.openwrt.org/doc/howto/o...mware.generate
Or just look at the makefile in it to see what is written into it.

Also the make -j1 V=s I just ran now finished without errors using the same exact .config file it failed with half an hour ago.

This is "normal". The process of pulling down stuff to compile follows dependencies if run as a single process, may or may not (usually not) follow dependencies when run as multiple processes.

I already shouted at them since I have a powerful multicore that I'd like to also use for compilation, their answer was "naah, I'm too lazy to do so" (literally).
I shouted again, Torvalds-style.

[starshipeleven]

Listening to the discussions it sounds perfectly reasonable to fork. If leadership will not keep pace, it is the job of contributors to take up the reins in their stead.

Hopefully unlike ffmpeg vs libav the LEDE developers will merge back in and cooperate again once their grievances have been aired. And it is the onus of the OpenWRT heads to see them met - you cannot long survive having a half dozen or more frequent contributors drop your project because you have a lack of participation and engagement.

Knowing the OpenWRT guys, it's not likely to happen.

It's more likely that the more reasonable ones will just jump ship when LEDE becomes clearly better (more meat for the meat grinder) and call it a day.

Which would be a great example of Opensource saving the day.

[SystemCrasher]

At least this LEDE thing has gone for git since the very beginning. Very good move. OpenWRT is slowly migrating to git in most parts as well, BUT it being half way there for a while. Sorry, but SVN is retarded. And even if you slowly migrate to git, it implies centralized svnish workflows. But git isn't about being "better svn". It about totally changing workflow, making it convenient for everyone. I'm pretty sure sorting this crap out could bring 'em new devs.

Then, OpenWRT's security isn't best thing ever: once some milestone is released, it is a bummer. No security updates would hit any package, no matter what. So networked equipment runs with, say, heartbleed bug. Or maybe that wpa_supplicant which can be remotely pwnzored. There was quite some fancy CVEs, after all. It could be better than that, right? But it is not about OpenWRT unless you run trunk. Which could break in arbitrary ways. Doh. So using openwrt and keeping things secure at the same time is a daunting task.

Then I do not really get why they put packages like they do. Logically, they want just 4-5 repos: armel, armhf, mipsel and maybe something else. Basically like Debian does. Right now every kind of device gets own packages. But most programs do not care if this some certain kind of device or somethnig else. It makes process strange, inconvenient and wasteful. I.e. they have to build same packages multiple times for similar archs for no good reason.

So OpenWRT is a good thing. But it has got own issues, and most of them are long-standing.

[starshipeleven]

For those wanting to have a good read, you can see the logs of their first meetings from here

https://www.lede-project.org/meetings.html

It's about 6-8 guys for now.

If you gather enough info to identify them from the logs, please share.

[wigyori]

From: mbm <mbm@openwrt dot org>
To: [email protected], [email protected]

Dear OpenWrt community,

It is with a great amount of surprise that, like all of you, we read
about the announcement of the LEDE project yesterday, as there was no
prior announcement nor clues this would happen.

While we recognize the current OpenWrt project suffers from a number of
issues outlined by Jo-Philip, in each of the 5 bullet points, we do not
agree with the conclusions withdrawn, and even less so in deciding to
spin off the OpenWrt project in the first place as a way to fix the
project and its community. Also, the phrases such as a "reboot" are both
vague and misleading and the LEDE project failed to identify its true
nature. The LEDE announcement contains a number of very valid points
which we hoped we had an opportunity to discuss and attempt to fix, in a
public manner, before this more radical outcome. At this point, the
email as well as actions taken are very confusing to a lot of us.

OpenWrt is primarily developed by individuals who may have a day job
more or less related to the purpose or the technologies of the project,
but who strive to maintain OpenWrt as independent as possible from any
company, organization or interest group, thus maintaining its own
infrastructure (website, forums, mailing-lists, bugtracker...), which
has been usually at the heart of all debates.

We do acknowledge there has been internal disagreements, on several
occasions about some directions of the project, about the release model,
the lack of testing, the centralized infrastructure, however, there have
been actual work going on under the hoods to solve things one step at a
time, starting with a more decentralized infrastructure, which was
discussed with the LEDE developers as well.

At this point, we do not have much to offer to the LEDE developers but
to encourage them to publicly discuss on
[email protected] the different items we should all be
fixing together, and avoid spinning off so that all decisions can be
taken with the community's involvement, and accountability and
transparency can rule us as one community.

As a user, developer, contributor, or just community member, whatever
choice you make, keep the choice that matters to you: the ability to
utilize superior quality open source software to power whatever embedded
device that matters to you!

We would like to stress that we do want to have an open discussion and
resolve matters at hand. Our goal is to work with all parties who can
and want to contribute to OpenWrt, including the LEDE team.

Sincerely,
Your OpenWrt team

[starshipeleven]

God, the above sounds like corporate crap.

I mean, are there also pages "intentionally left blank" in the mail?

[dave773]

Main developer and some others have their companies (as DD-WRT) which they advertise more and is more important to them than OpenWRT.

You are aware that these are the people now in LEDE.

[SystemCrasher]

the different items we should all be fixing together,

At very least these wrenches have to fix is the fact their web server returns 403 for half of config files in "trunk". Sure, who gives a fuck about ability of those trying to build own images to take a look on known-good configs, right? it has been like this for whopping year. Or maybe two. If not three, I do not remember.

Wanna report bug? Go suddenly figure out their bug tracker is brain-damaged as well. It would not let me to register, so I can't track bugs I've reported. However it allows anonymous bug reports and comments. Unfortunately it leads to loads of spam. So devs are rarely reading bug tracker. Good luck to find your bug report month later either. Doh.

If miracle happens and some OpenWRT dev stumbles on bug report, it is 99.9% chance it gonna to be WONTFIX anyway. It is very rare to see OpenWRT devs fixing bugs, even most pressing ones. So it continues crippled, half-working, full of known vulns. Even releases have truly glaring bugs, fucking up day to day usage for many users. Who cares?

And if someone is about shady agreements and corporations, OpenWRT devs pioneered in this. Example: TP-Link 1043 can't use channels 12/13 in Europe/China/etc if running OpenWRT. Good luck to connect AP running on chan 12/13, etc. OpenWRT smartasses deleted regdb/CRDA and now there is no reasonable way to do something about it at all. If one is truly fed up with glaring Qualcomm-Atheros BS and being really persistent, devs could unofficialy advise to "borrow" battle cruiser they've just made, covertly stored in hangar, enter low-earth orbit and shoot high-energy beams on everything daring to oppose. I mean, there is epic wireless hack which would allow one to ignore ALL limits, up to disregarding these freakin' radars, ingnoring all power constraints alltogether and other stupid stuff, so you can just shoot high-energy beams anywhere. But hey, if I want to connect AP on chan 12, it is not necessarily implies I want to jam some stupid radars or shun low power APs around. Somehow one either to be crippled "pacifist" or ride high-profile battle cruiser. No other options provided.

Whatever, but 99.9% of time attempt to report bug to OpenWRT is futile, nobody cares, or nobody would spot you've reported it, or current policies (or devs views, or shady agreements with corporations) would bring it to halt, despite of planet-wide impact or other unimportant crap. That's how it works. Realistically, running OpenWRT is great. If you could afford dev-like & system integration expertise.