Librem 5 Developer Kit's Mainline Kernel Support Hits 12th Patch Revision

[tg--]

I'm a backer and I fully expect them to not hit the 2019Q3 mark. In fact, I find it highly unlikely and 2020Q1 seems a lot more reasonable.
Still, I think they are trying to do it right and made a lot of choices that are on the right track; thus I consider their effort the most worthwhile around and a project worth supporting.

However, they are clearly spending too little effort on the hardware and driver side, and trying to do too much high-level software.
This will cause more delays and I'm not at all sure if they can put out their intended software platform along with the hardware without delaying the latter too much.

However, if they don't manage to establish their software-stack right from the get go, we will once again see a lot of fragmentation issues.
The KDE/Qt guys are already working on a largely incompatible KDE based system and will push hard against the somewhat gnomish stack of Purism, which will make a unified system a lot harder to establish.

I think this effort will make it pretty much impossible for the Librem to become an actual Android alternative, despite good intentions in both camps.
It would be crucial to find a common ground, i. e. a middleware that both camps would be happy with, and allow both softwarestacks to coexist without fragmenting the whole platform; but KDE has its own middleware already and I don't see them moving to something else - nor the other camps to move to KDEs stack.

[ElectricPrism]

wtf is up with the ad

[L_A_G]

Why not both?

Yes, why should you ever try to keep what you've promised? It's not like people won't judge your future promises based on how you've kept your promises in the past...

Also, is that even bad? They are working on it and they are showing progress. Why ship a half-baked shitshow for the sake of not missing a "launch date", people has waited until now, they can wait a few more months.

You'd have a very good point if this wouldn't be another delay of several months on top of multiple previous ones. Originally this thing was supposed to ship late last year, then it was supposed to ship in April this year (i.e last month), now it's supposed to ship within the July-September time frame and you're suggesting another delay that would inevitably push shipping into next year due to how busy contract manufacturers are at the end of the year on account of the Christmas season.

Don't get me wrong, particularly in terms of software this thing really doesn't look like it's ready to ship to end users, with or without an out-of-tree kernel. It's pretty clear that their approach of getting custom dev boards done and only then starting the development of the stack proper has delayed progress on said stack pretty significantly. However delaying it even further will only slow down the development of the software stack, i.e it's biggest weakness at the moment, and it obviously needs as many devs contributing patches and bug reports as possible.

Thus if I could make a suggestion it would be for Librem to offer backers a choice between waiting for the "consumer ready" stack to be ready (which I personally doubt will be ready until early next year) and a "developer edition" device with an out-of-tree kernel and the software stack in whatever state they can get it to within the promised Q3 time frame.

[starshipeleven]

Yes, why should you ever try to keep what you've promised? It's not like people won't judge your future promises based on how you've kept your promises in the past...

Hey, easy there. I didn't say that they should break ALL promises. My point is only that the promises about software and hardware quality are orders of magnitude more important than promises about the release date, and that breaking their release date (again) isn't a major problem imho.

[L_A_G]

...

As I said, just breaking a release date promise once isn't really an issue, but when you do it repeatedly like Librem has already done it's definitely going to be an issue going forward. If they're going to delay it any more it's inevitably going to mean a delay into next year because of how busy contract manufacturers get by the end of the year.

Your personal feeling doesn't really matter when we're talking about something being distributed to a wider audience and which is inevitably going to contain a significant contingent of people with a very different attitude from yours regardless of what it is. Not only that, with the delay in getting development on the software stack into full swing they really do need as many devices in developers' hands and the only way to do that within this year is to at least get a limited "dev edition"-release out the door before contract manufacturers get too slammed with production for the Christmas season.

This is obviously not an ideal outcome by any stretch of the imagination, but with the hardware practically done and software stack needing a LOT of work, I honestly don't see any better alternatives.

[starshipeleven]

As I said, just breaking a release date promise once isn't really an issue, but when you do it repeatedly like Librem has already done it's definitely going to be an issue going forward.

I'm still saying it's much less of an issue that you think.

If they're going to delay it any more

They got their money already, and bulk of their hardware is a SoM. What stopped them from ordering the hardware until now? Are they still developing their hardware? If that's the case, I have bad news for you.

Your personal feeling doesn't really matter

It's not my personal feeling, it's what I've observed in other crowdfunded projects.
People in the crowdfunding sphere are more forgiving than the average customer buying a known finished product. Those that aren't get weeded out of pretty quickly by delays and scams and swear they will NEVER EVER crowdfund again.

Hardware projects like this commonly miss the mark by 6 months or a year when they aren't provided by a company that has experience in the field.

Not only that, with the delay in getting development on the software stack into full swing they really do need as many devices in developers' hands and the only way to do that within this year is to at least get a limited "dev edition"-release out the door before contract manufacturers get too slammed with production for the Christmas season.

Yes I agree with this. I was talking of devices for consumers above, not devkits.

[L_A_G]

I'm still saying it's much less of an issue that you think.

The thing about "Well that's how I feel" type arguments is they're essentially worthless. You're just one individual and there's literally 1000s of backers over the 3500 devices they've pre-sold so far.

They got their money already, and bulk of their hardware is a SoM. What stopped them from ordering the hardware until now? Are they still developing their hardware? If that's the case, I have bad news for you.

The bulk of the hardware is definitely not an SoC, it's probably less than 20% of the whole parts budget for a device. To produce 1000s of those devices is going to require you to source quite a few components and most bigger component vendors generally deal in bulk orders several times that so you're probably going to have to work trough distributors who obviously aren't going to have that amount of everything in stock.

After that you need to have the custom PCBs you've designed for it manufactured, then you're going to need to get the PCBs populated with the components and finally have the whole thing assembled at another factory. There are companies who will take your PCB designs and your components and then give you back fully populated PCBs ready to be sent off to be assmebled into devices. However there are companies that specialize in one or the other and with the manufacturing rush for the Christmas season fast approaching it's not a given that any company that does both is going to have the capacity to spare.

When it comes to the design of the actual final hardware, if their most recent hardware update is anything to go by then the final hardware should either be done by now or then they're putting on the final touches as we speak. Unless there's some serious issues they've only recently discovered then the hardware has no reason to not be ready for production so that it'll start shipping to end users before Q3 is out.

It's not my personal feeling, it's what I've observed in other crowdfunded projects. People in the crowdfunding sphere are more forgiving than the average customer buying a known finished product. Those that aren't get weeded out of pretty quickly by delays and scams and swear they will NEVER EVER crowdfund again.

Again, this is not the first time they've delayed this and if you want a good example of hardware-based crowdfunding projects being delayed to the point of the backers eventually losing trust in the project just look at the ZX Spectrum Vega+. The backers finally lost patience about 8 months in and a year in even Indiegogo had to get involved to force them to get the devices out the door.

[starshipeleven]

The thing about "Well that's how I feel" type arguments is they're essentially worthless. You're just one individual and there's literally 1000s of backers over the 3500 devices they've pre-sold so far.

Not only that. I'm also one that is NOT paying for silly promises, especially for noob companies like Purism. I know hardware and firmware enough to have forseen their "delays".

I did pay for crowdfunding about the La Frite https://www.kickstarter.com/projects...n-source-fries and the Helios4 Nas https://kobol.io/helios4/ but there was a "collect money to actually build a batch of hardware".

What I said isn't my personal opinion. It's how I've seen things go.

The bulk of the hardware is definitely not an SoC,

I said SoM. System on a Module. It's a mini-motherboard with RAM, SoC, Flash and other stuff on board. It exists because buying one of these for 100$ apiece (or less) is ORDERS OF MAGNITUDE cheaper than designing your own board with the same capabilities.

This is a SoM for the hardware they are using, and a phone-compatible form factor. https://www.compulab.com/products/co...dule-computer/
There are others. https://www.intrinsyc.com/computing-...on-module-som/
And others. https://www.solid-run.com/nxp-family/imx8-som/

And this one is what they are actually using in their devkit (it's the smaller board with the blue heatsink you see mounted on the bigger board) but it is a tad too large for a phone https://emcraft.com/products/868
(see their devkit https://puri.sm/posts/librem5-2018-11-hardware-report/ )

If they are not using a SoM they are fucked. Simply and throughly fucked, you can start laughing at them now as they WILL crash and burn, no other possible outcomes. Hardware design at this level of miniaturization and integration is NOT something you "fake it until you make it".

if you want a good example of hardware-based crowdfunding projects being delayed to the point of the backers eventually losing trust in the project just look at the ZX Spectrum Vega+.

That's different, the ZX Spectrum Vega+ was a completely opaque cash grab, where none knew what was happening beyond some generic statements every now and then. It was hard to even get pictures of products from them.

With this device they are posting regularly about their progress and you can see their work in the source code repos.

[starshipeleven]

aaand unapproved post for L_A_G above