Wow… great work, development team!
A great report again, detailed and informative. Very nice to see all the stability improvements. And only four blockers for the final release left, at the moment of writing. Thanks to all who contributed to this!
It contains some really nice improvements. A lot of the Mesa work that dos backported was done by Guido GĂĽnther, Christian Gmeiner and especially Lucas Stach. Kudos to all of them (Purism Dev-Team, too)!
I must say, the bar should be a little higher than “the best PureOS release yet” - as true as that may be, there ain’t that many ![]()
I think we can find some better comparison to make.
A minor detail to mention that I didn’t look into but got curious afterwards: is there an “enterprise feature” (or for the family uber-nerd) coming to PureOS Dawn (or is it a phosh thing) that allows for an organization level admin code - or did I misunderstand what I saw in the settings?
I’m so looking forwards a final release and update of all the packages.
Gnu crimson need “Lean and Mean”
0x0
But etnaviv is still 22.3.6 instead of the last 24.2.4 so we’re still using a 3 years old etnaviv. In fact my L5 is still jerky although updated
These are backports for a PureOS version we probably do only (hopefully) run for some months. How much more work should dos put into a temporary build? Every backport costs time that pushs Dawn to a later release date. As more modern the version as more troubles to backport the software, because all dependencies still have to work.
I think we should wait with such complains/requests until PureOS Dawn is released. At that point we can request newer features of some packages. But even then it only makes sense if Librem 5 gets important improvements. that will make the overall user experience better. It makes no sense to backport packages if they don’t provide a benefit for average people.
I thought etnaviv (smoothness) were one of the most important for average users ![]()
My apologies ![]()
But not every version brings more important smoothness (and some can even bring new bugs). I would hope to get GLES3 backport on Dawn, once it will be released.
But at the end the devs have to make the decision (they also see problems we don’t see).
It’s the time of the month to start to wonder, if the blog is skipping (again, which didn’t use to be a thing)? Or never mind about that: are we finally getting something real or is it still “forward momentum”…? Milestones seem to be done - what next, when next, is there next?
JR-Fi,
the last month was a mess through anthropic myth A.I. to all open source projects.
It done a great rift in technology, special to the open source because it read and learned from this in the first place. But i think purism got no special access kid because it was to small. However they have to merge all the patches and Linux Kernel exploits too, an there are many.. i have no patched Copy Failed 2 patch on my L5 right know so .. yes it will be next.
But about Milestones. We have a future, because we are open and connected to each other and to the free LLMs out there, caring about privacy.
dos wrote previously that the Librem 5 is not affected by Copy Fail.
However that CVE appears to have spawned 3 other related CVEs and I don’t know what the status of those are.
Dirty Frag (aka Copy Fail 2), Fragnesia and ssh-keysign-pwn are all fixed in the Crimson’s L5 kernel (Saturday’s 6.12.89pureos1).
6.6.139pureos1, currently in byzantium-updates-proposed, fixes them all too except some aspects of Fragnesia that haven’t been fixed upstream either so far (crimson applies an extra patch for that that hasn’t been merged upstream yet).
Thank you and all the one for updating this. Like Linus said there are many duplicate reporters for the new many folks who only search/ask some A.I. and reported it but not send in a patch. today.
From users view we have to take care and patch and i saw only a 6.12 Kernel. now i am on point 89 too. I will support you.
Just dropping in to say thank you for the heads up on the dev update. I had to go back to Android simply because of unreliable call related issues. Looking at their update:
Calls, the calling application, now prevents starting a second call while one is in progress. Calls doesn’t support multiple calls, so if a second call ever did occur, it would just hang it up again. This isn’t normally possible, but this robustness fix ensures that we don’t dial a number inadvertently if it does happen due to a bug.
Occasionally, we’ve seen the modem’s communication interface to the phone stop responding. While the cause for this is not clear yet, and likely involves the modem’s firmware, we improved ModemManager to reset the interface if it happens to maintain connectivity.
A common “call quality” issue is actually caused by microphone input volume. If it’s too low, the other party can’t hear you. If it’s too high, the other party will hear distorted audio and might receive excessive background noise. It is also hard to adjust, as the setting is part of the detailed audio settings, and you can’t hear the result while changing it. We’re now setting a reasonable default level on boot to ensure that calls work.
Finally, we backported several fixes for ModemManager and notification improvements for Chatty, which improve the overall phone experience on the Librem 5.
AMAZING. I really want to get back to daily driving my Librem 5. This is looking promising!
I also did not realize there were dev subscriptions available, will be signing up to help.
If any devs are reading this, please forward my many thanks. Looking forward to a free future.
Oh, also, everyone knows but so it is said and recognied; HARDWARE if fricking hard.
You can build free hardware all you want, but then you have the software and social problems. You have to build the software for the hardware and you have to build the social infra for the software.
And then, each of the 3 (hardware, software, social) have bugs, issues and maintenance and care requirements. It is a BLOODY DIFFICULT thing get done at all never mind get done right.
However, if Purism can get the L5 calls and text/mms stabilized so it is rock solid, the rest is cake and can build from there.