hi @lanodan, you jus mentioned #openwrt.... i recently formulated a spray to deal with DSL modems, none of which are able to be liberated with foss firmware. :(
@lanodan there are apparently proprietary blobs that openwrt allow (especially for the modems). They apparently allowed the blobs for convenience otherwise we'd have to go on a global #strike or something. Maybe we ought to.... is a pretty sorry situation really.
This is exactly why I don't care about FSF opinions. Because almost no system ever is 100% full source, meanwhile it's trivial to document so users can make choices in their best interests and developers clearly know where the problems are.
But I'm very critical of the FSF, in theory they are strict and they love presenting as such. But then in practice they aren't, which they don't even try to document out. To me it's like if you'd have a big vegan organisation fully approving vegetarian meals without any disclaimers.
And for routers… to me best one you can get from a freedom perspective is something like a PC Engine APU, which is brickless even for normal users since you can physically override the bios chip by plugging in another one, and since it's an x86-pc you can use pretty much whatever OS you want. Compare that with voluntarily scuffed hardware where you're stuck with a specially made OS and pray it won't brick at each update (and pray the soldered NAND/eMMC won't burn too badly).
@lanodan > apparently i trawled through the openwrt forums not long ago and they admit to it.... the non-free issue is not mere opinion sadly.
> meanwhile it's trivial to document so users can make choices in their best interests and developers clearly know where the problems are
i don't know what is meant here, when a vendor say proprietary blobs it pretty much means i have no idea if the thing is really secure, we are entering into a time where data is important and could mean life or death for really anyone. i just don't see how to move forward until devices are liberated. You could ask "what can a bad modem really do?" but they could do all sort of nasty things. If you asked me 15 years ago i wouldnt have cared as much, but today, i have to care. I wish i could un-care about our tools not being free, i just can't. the number of attacks against regular people are so high these days, i talk to totally random people and smart, careful people are getting into trouble, its clear that safer tools are needed.
thanks for your thoughts on the topic anyway, i wont bother u with libre-centric stuff now that i kno is not your realm.
@lanodan@monnier@frogzone "What if OpenWRT but with GNU Linux-libre instead of proprietary Linux and no proprietary software." - libreCMC.
The cognitive dissonance is incredible - the FSF is too strict for saying that subjecting the user to proprietary peripheral software is unacceptable, but is not strict enough by not insisting that 100% source bootstrap is always immediately available (a few binaries aren't a problem if you know every binary was built from 100% free software, although full source bootstrap without manual steps would be an enhancement)?
@Suiseiseki@lanodan@frogzone Oh, I think this is a "technical" issue: the FSF has not yet understood that bootstrapping is a problem. I suspect that part of the reason is that you can modify/replace the bootstrap blobs (tho you have to use those blobs in the process), so if you assume the seeds are honest, it's not as bad as firmware-blobs which you can't modify. [ As a data point: until version 20, even building Emacs required blobs (in the form of precompiled .elc files). I fixed that in Emacs-21, but back then I did not understand what it meant: I did it only to make it easier to compile after a cvs checkout🙂 ]
@Suiseiseki@monnier@frogzone I never said in this thread the the FSF is too strict, and my point with FSF perceived strictness has always been that they should at least somewhat document their claims when they make some, and at least be consistent. And to me strictness should be based on rules, not "I'll not accept blobs from $x but I'll accept blobs from $y", blobs are unacceptable at least by default.
And while LibreCMC is nice, it's a special system to support hardware that's made difficult to hack on, quite like modern Apple hardware or Android smartphones if you want something painfully obvious. At least to me it doesn't even matter what software you run on them because they're less computer than a programmable calculator.
@lanodan@Suiseiseki@frogzone The FSF's position is based on rules, which are the 4 freedoms. In this sense bootstrap-blobs can be studied/redistributed/modified (tho only if they were produced honestly, which can be hard/impossible to verify). That's the difference with your usual firmware blobs for which you don't even have anything that claims to be its source code. Remember: the first "F" in "FSF" is for "Freedom", not "Fecurity". [ As for OpenWRT/LibreCMC/others: I run OpenWRT rather than LibreCMC because of the DSL firmware. And I run OpenWRT rather than Debian because that one machine-with-the-modem is a bit too limited to run Debian comfortably. I don't know of a machine-with-the-modem that can run a "normal" GNU/Linux distro. I suspect in the future that machine will be replaced by a 100% proprietary one 😞 ]
@monnier@lanodan@frogzone >the FSF has not yet understood that bootstrapping is a problem. GNU did the bootstrapping process decades ago and really enough changes to the software have been made that any backdoors a proprietary compiler inserted couldn't possibly have survived (gcc output has been inspected by many, many people and any insertion of backdoors would have been noticed).
The FSF understands well that freedom is still lacking and is currently far more important than full source bootstrap, although such is a nice to have and won't stop anyone from carrying out such.
>even building Emacs required blobs Past the initial bootstrapping it never did - considering that a blob is a piece of proprietary software.
Precompiled free software .elc files that correspond exactly with provided free software source code isn't a freedom issue.
@lanodan@monnier@frogzone >it's a special system to support hardware that's made difficult to hack on You just ssh in and then you can do whatever computation you want, maybe plugging in a flash drive to get more storage or swap if needed?
>quite like modern Apple hardware or Android smartphones if you want something painfully obvious. Unlike modern apple hardware or Android smartphones, such devices run with 100% free software and do not contain any digital handcuffs to restrict what you run.
>it doesn't even matter what software you run on them because they're less computer than a programmable calculator. Well, there it is - it doesn't matter if software running on a computer that happens to be most useful for routing is free or proprietary eh?
They're computers, although they only really have a inconvenient UART shell for display, although that doesn't matter as you can just ssh in, install GNU bash and do whatever computation you want.