Pandora should leave Angstrom

[ShleeDragon]

3. non-public sources for SGX and DSP, that we are not allowed to redistribute, but can be retrieved if you register at TI and pass their software export screening thing.

Are you still getting the pandora patches merged upstream? how much patches are required for the current 3.2 stable to compiling for pandora.. Can you give me a list of all of your patches or should I just clone your Git?


also, the SGX/DSP.. Are these modules/userspace binaries? Id assume you can compile a kernel and package up the matching modules/userspace for this sucker.


I'm in the process of trying to setup a archlinux build + public repo system to start building PKGBUILD if required and looking at the requirements...

 

[ibisum]

It _does suck_ to start building OE, though.


..


We do have a build server that spits out all the ipk's and such; thats where notaz and ED run builds

So all I'm asking is that this somehow be made a little more public to those who wish to contribute, and that a tiny bit of effort be put out there to make it possible for interested parties - such as, but not limited to, myself - to start participating in pushing things forward. Not all firmware-OS contributions have to be kernel patching/configs/fixing - some things could be done in user-land to make the default firmware more friendly (such as, oh a random idea: finally having a proper nub-calibration stage integrated in the boot process, with GUI and all..)

I'm not really sure it will help, if you can't bother to try checking out that git and trying a build (haven't seen any report from you about that), will you bother to do any firmware work?

I value your technical contributions Notaz, but somethings I think you can be a real asshole. I have checked out the git, I have attempted a full build, and I have run into problems that leads me to conclude that the leg-work you have done is more valuable than you seem to think it is. A VM would solve this.


My point was: make it easier to contribute by reducing the redundant work required to keep up with you, and probably you'll get more contributions. Your response "if you can't be bothered doing all the work I've already done, what work will you do in the future?" Yeah, the point is: why should I have to do all that work if you've already done it, already?


I'm trying to contribute by not having to play catchup with you with the build system, and all its intricacies. You've got tons of stuff on your plate that is local, pretty much exclusively, to your own system and which isn't out there for other firmware-OS devs to stand on. You can't tell me that its going to be 'easy' to catch up with you when I have to build everything from scratch, get the patches integrated, "telepathically know what notaz is up to", and all that.


Lone-wolf operations may be glorious, but its not teamwork. If you're going to lead the way, tape a path so others can easily follow you, catch up, and scout ahead of you if needed. I only know of a couple of guys, who seem to have time for nothing but Pandora, who have managed to keep up somehow - my valuable hacking time, spent on Pandora, has been in userland for the most part. I'd freakin' love to contribute more to the base OS situation, but .. 12 hours of work to do before I can even find out how fucked things are, first? Not gonna happen, alas ..

Well, Notaz. Thanks again for all your efforts. You are the one who knows the system and it's benefits/problems the most. Do you have something like a roadmap? It seems you like to keep your current project a secret, but it might actually help to team up if we knew your priorities.

New people coming to this project shouldn't have to use telepathy to find out whats up. A bug-tracker, a build-system, a repo with sensible configurations, and off we go. Imagine we get an amazing kernel hacker who can solve all of notaz' unsolved issues come to the Pandora project next week - what will they have to do to catch up and overtake notaz? Telepathy and .. umm .. reading all of notaz posts for the last year, instead of 'git log' or some other such sensible approach ..

 

[notaz]

Are you still getting the pandora patches merged upstream? how much patches are required for the current 3.2 stable to compiling for pandora.. Can you give me a list of all of your patches or should I just clone your Git?

Mainline boots on pandora for a while now.

also, the SGX/DSP.. Are these modules/userspace binaries? Id assume you can compile a kernel and package up the matching modules/userspace for this sucker.

Yes it's userspace.

So all I'm asking is that this somehow be made a little more public to those who wish to contribute, and that a tiny bit of effort be put out there to make it possible for interested parties - such as, but not limited to, myself - to start participating in pushing things forward. Not all firmware-OS contributions have to be kernel patching/configs/fixing - some things could be done in user-land to make the default firmware more friendly

You want public build server access? What if someone breaks it, intentionally or not? It's enough to remove one directory and you lose the build because of OE rot problem I detailed about before. And then it's not only you, but also we don't have a working build server, and there are no firmware updates any more until it can be revived, which would take lots of valuable time. Ok ED has backups, but do you really want to create more work for him?

(such as, oh a random idea: finally having a proper nub-calibration stage integrated in the boot process, with GUI and all..)

That's not even possible, software has no control of that, it's all in nub firmware.

I'm not really sure it will help, if you can't bother to try checking out that git and trying a build (haven't seen any report from you about that), will you bother to do any firmware work?

I value your technical contributions Notaz, but somethings I think you can be a real asshole. I have checked out the git, I have attempted a full build, and I have run into problems that leads me to conclude that the leg-work you have done is more valuable than you seem to think it is. A VM would solve this.

And what was the exact error you got, may I ask?

My point was: make it easier to contribute by reducing the redundant work required to keep up with you, and probably you'll get more contributions. Your response "if you can't be bothered doing all the work I've already done, what work will you do in the future?" Yeah, the point is: why should I have to do all that work if you've already done it, already?

Because that work now doesn't mean anything, yet again because of same OE rot problem, I would have to do it all again from scratch, just like you. Why do you think your time is more valuable than mine?

You've got tons of stuff on your plate that is local, pretty much exclusively, to your own system and which isn't out there for other firmware-OS devs to stand on. You can't tell me that its going to be 'easy' to catch up with you when I have to build everything from scratch, get the patches integrated, "telepathically know what notaz is up to", and all that.

What are you talking about? What system? Everything is pushed to gits, build server can only fetch changes from public gits. There are people successfully building this stuff.

New people coming to this project shouldn't have to use telepathy to find out whats up. A bug-tracker, a build-system, a repo with sensible configurations, and off we go. Imagine we get an amazing kernel hacker who can solve all of notaz' unsolved issues come to the Pandora project next week - what will they have to do to catch up and overtake notaz? Telepathy and .. umm .. reading all of notaz posts for the last year, instead of 'git log' or some other such sensible approach ..

bug-tracker: doesn't current one work?


build-system: sure would be cool, perhaps something web based, but who is going to develop that? Giving access to main one is not an option, read above.


repo: again, no one to do it, with lots of empty promises from yourself.


telepathy: what are you talking about again, guess this is what you call telepathy then?:


http://git.openpando...g;h=pandora-3.2

 

[ibisum]

That's not even possible, software has no control of that, it's all in nub firmware.

Umm .. its an app that shows the position of the controls, a little GUI that shows the areas not yet 'moved to' yet by the user, and the user 'scrubs the area out' - thus calibrating the nubs the first time it boots.

You know, the "calibrate the nubs by moving them around all over the place after you first turn on your Pandora", calibration step. Not some software magic - a simple app that prompts the user to calibrate.

Everything is pushed to gits, build server can only fetch changes from public gits. There are people successfully building this stuff.

Great! I guess I have to RTFM more often to find out how this is done, and how to contribute to it.

build-system: sure would be cool, perhaps something web based, but who is going to develop that? Giving access to main one is not an option, read above.

So we have a staging tree, and a production tree, and the build-server builds from staging for contributors, and production for master firmware producers, such as yourself.

Why do you think your time is more valuable than mine?

I don't make any such distinction but its pretty clear you have more time to work on Open Pandora than I do. Thems the breaks, we both have day jobs and other lives. I'm only asking that we get some sort of system in place to welcome contribution, get past the barriers to entry for others, and open the firmware development/build process up a little by providing some tools. I still think a VM with everything onboard would be very useful, especially if it was able to reproduce the production firmware OS images with ease .. anyway enough chuff and hot-air from me, I'm going back to user-land stuff where I can spend my limited time with some degree of result ..

 

[pmprog]

also, the SGX/DSP.. Are these modules/userspace binaries? Id assume you can compile a kernel and package up the matching modules/userspace for this sucker.

Yes it's userspace.

As somebody interested in starting to help out with the firmware, this is probably going to sound like a totally thick question - Is userspace like "Ring 3" - ie, does not run under the Kernel? if so, does that mean if I were to try and compile the firmware, all I would need are the binaries off the current Pandora firmware, as opposed to the source?


Not that I'm bothered about trying to register to get access to the source, just thought if I didn't need to, I could just get it to work off the binaries. It would also make any VM disk image I come up with completely legit

I still think a VM with everything onboard would be very useful, especially if it was able to reproduce the production firmware OS images with ease

I think this totally beats a public build server; as notaz said, only takes one person to cock it up. Where-as if you have a VM image, you can just re-download or copy the disk image at various points.


Collector Plus is at the top of my personal dev priorities, but I shall try and look at starting off this VM image as and when I can.

 

[ShleeDragon]

This thread is kind of amazing.. unless you have a basic understanding of the Linux environment - Id recommend staying out and leave the heavy lifting and brain thinking to Notaz.


I think the problem is, some of us want to learn to compile an OS using the pandora (followed by trying to help improve the default OOB experience) and the other half have some kind of problem with the current OS firmware and they think the problem is related to not having enough chefs in the kitchen,


Notaz will obviously continue doing whatever he thinks is right for the project.. don't pretend for a moment he doesn't have the projects best intentions in mind.


My two cents is: I believe the pandora is more tool than toy - and secondly would like to see a more bleeding edge OS running on the unit. While for most users, stability on the OS and the ability to run all of the PNDs without fault is obviously more important and this has lead to the standard 'stable' point firmware releases we've seen...


---


Update: Id recommend a book called 'The linux programming interface'. it's expensive, but you'll learn the ins/outs of the linux kernel pretty well. Kernelspace and userspace generally means the allocated memory and IO & memory limitations placed on the system. The kernel can read/write directly to the hardware and all of the memory. The userspace has to ask the kernel to perform changes to anything and can generally only access information within the allocated userspace.. This can mean a binary like 'rm' is userspace because it asks the kernel to remove the file and the kernel does all of the heavy abstraction to talk to the filesystem/disk to actually delete the file.

 

[ibisum]

FWIW, I'm not a Linux kernel newbie; I developed kernel modules and extensions to the Linux kernel, as well as user-space apps, for a major provider of realtime, safety-critical systems here in Europe for 5 years until moving into the mobile space. My Linux kernel work is guarding rail transportation systems in 38 countries. I'm a SIL-4 rated programmer, and have been using and developing for Linux professionally since Linus announced it on the minix-list. I'm not able to contribute to the OpenPandora due to time constraints, and my desire for a more organized firmware-OS development production line is derived from this experience, and the desire to help as much as possible. At this point though, I believe you have a valid point: we should just let Notaz sort it out, along with anyone else who has the time to do the heavy lifting required. In the meantime, I'll just stay focused on user-space and continue to gripe about the situation at any chance I'm given.

 

[pmprog]

This thread is kind of amazing.. unless you have a basic understanding of the Linux environment - Id recommend staying out and leave the heavy lifting and brain thinking to Notaz.

I find that pretty insulting TBH. Sure , I'm new to Linux, but I'm not expecting to write a new kernel, just do a bit of leg work to assist; and possibly get myself into a position to help with bigger things down the line. You're seeming to prefer forcing everything through an artificial bottleneck rather than embracing a community.


Maybe Notaz and co don't want help, that's fine, I have plenty of other projects in my list.

 

[Linux-SWAT]

That's not the point, of course everyone is happy when someone gives help.

 

[ShleeDragon]

Sorry if I seemed to be hitting your 'ad hominem' bone. I never said, if you wanted to help - too bad.. I'm not anyone of any importance... I do remember saying, this thread is filled with people who want to learn, but are stuck and people who are willing to help and are stuck.

 

[pmprog]

Sorry, it just sounded that way when I read it.


I guess I'm not really in a position to start helping out yet anyway, so I'll probably retreat to a dark corner until I'm ready

 

[notaz]

If anyone is stuck on anything, they should talk. This thread is an example how that works. From my camp it looks like only a few are even trying to do anything.

 

[sebt3]

From my camp it looks like only a few are even trying to do anything.

+1


Also, the best way to get started is the irc channel

 

[klaxonn]

Also, the best way to get started is the irc channel

Not so sure about that, some people don't really motivate to do anything. At least, it was the case for me.


The wiki have enough information to start.

 

[sebt3]

Not so sure about that, some people don't really motivate to do anything. At least, it was the case for me.

The wiki have enough information to start.

So you hitted aTc ? Dont you ? He is probably the most helpfull guy I know.... once you know how to handle his way

 

[Grench]

If anyone is stuck on anything, they should talk. This thread is an example how that works. From my camp it looks like only a few are even trying to do anything.

There are what? @ 3000 Pandora's in the wild now?


Assuming 1% do actual development work - that's about 30 people.


Of those 10% of those work on the system while 90% work on applications - that's 3.


Not overly surprising really - not likely a 'contribution' answer.


On the bright side, if we can double the number of consoles, 'in the wild,' the number of developers should pretty well double as well. They're going to need a trail of bread crumbs to get started though.


I'm seeing a few people say they want to help - but don't know where to start to understand what has already been done.


Maybe someone who knows how it all works should leave a clear set of, 'What you need to read to understand where to start on Pandora OS and systems development' instructions? Not writing the book, but writing a post with links to the first 80% of the information that sets them down the right path? For better or worse, if this is going to be done - it sounds like Notaz is the authority who would need to do it. If he's going to get some help, though, that is possibly the best place to start.


Just a thought.

 

[skeezix]

Perhaps a VM would be doable, given some conditions..


1) No SGX or the like at all; ie: coudl not produce a real final firmware, but could produce a mostly working one.


---> possibly some of the SG stuff could be yanked from the actual pandora; ie: the wiki could include some notes on how to get that stuff and put it into the VM perhaps, though not really necessary imho


2) Out of date packages; running a full bitbake etc woudl blow up, maybe it could be set up to build as is, so that only pandora vendor overlay could be tweaked and the rest woudl 'just work'


So the process woudl be then to get the VM, run a git pull to get the latest bits, and then see if it makes 'as is'.


The problem is.. the recipies will have all wrong version numbers and such, so the normal baking won't work; I'm not so familiar with the processd to know but perhaps theres an option to say 'just build with local versions, donmt' try to pull the one in the recipies..' or something.


Now, a lot of this is moot, since.. libpnd for instance (including the daemons) can build on any build environment pretty easily; it has no crazy dependancies. The kernel is tricky though, need a good setup to build a working kernel. So it depends on the person's goals.. if they just want ot hack on the boot sequence or userland scripts, thats all ther ein git right now, and easy to hack on.


A VM for building a full firmware could work, but I wonder how many woudl jump on it.


That said..


X) Givne one of the existing dev VMs (sebt3's? yactfreeau or whatever), coudl co-opt that; I mean, just pull git's and go, right?


Y) Another problem with VMs is .. do you have any idea how much disk space is needed to build a firmware? That VM image is going to _grow big!_


torpor -- anyway, yeah, so you want to help.. great. Can you build the VM?


We are all quite treceptive to ideas as you know; theres IRC, theres the official mailing list, theres the forums here; we've tried to document how to go about it, though maybe some of those wiki pages are hard ot find who knows. But folks need to jump in, theres not enough critical mass to make it that easy.


Maybe once we get another 10,000 units out, and more devs up, some of those guys can help with better docs, better build tools and experience, and more info sharing, etc.


jeff

 

[skeezix]

pmprog .. I have some experiments, but out of tune a bit right now; teethign babies, then March Break, and a good friend of mine just passed from cancer (too young, mid 30s :/), so I'm rather busy and bummed out at the same time. Thats all excuses, I shoudl get on it, but bummed out all the same Ah, to have the free time I had even 6mo ago


If you're open for testing, maybe I can post something for you in a week or two. (networking, wise.) (Also, need to start considerting overhaiuling the keybindings and layouts .. getting to be a lot of functionality in there, and its not well laid out for it; also, we probably have a better idfea of how thigns are used since it was started so could make it better tuned for those workflows .. at the expense of losing existing skins, mind.)


jeff

 

[captain_coconut]

FWIW, I'm not a Linux kernel newbie; I developed kernel modules and extensions to the Linux kernel, as well as user-space apps, for a major provider of realtime, safety-critical systems here in Europe for 5 years until moving into the mobile space. My Linux kernel work is guarding rail transportation systems in 38 countries. I'm a SIL-4 rated programmer, and have been using and developing for Linux professionally since Linus announced it on the minix-list. I'm not able to contribute to the OpenPandora due to time constraints, and my desire for a more organized firmware-OS development production line is derived from this experience, and the desire to help as much as possible. At this point though, I believe you have a valid point: we should just let Notaz sort it out, along with anyone else who has the time to do the heavy lifting required. In the meantime, I'll just stay focused on user-space and continue to gripe about the situation at any chance I'm given.

Speaking as someone who has never done any serious work on the kernel, I'm not sure where you got stuck. If you want to start hacking on the pandora kernel there aren't that many steps to get an build environment going.

git clone git://git.openpandora.org/pandora-oe-environment.git

cd pandora-oe-environment

. ./op-env-oe.sh

./initial-setup.sh

bitbake omap3-pandora-kernel

 

[pmprog]

Y) Another problem with VMs is .. do you have any idea how much disk space is needed to build a firmware? That VM image is going to _grow big!_

I don't, but there's things you can do to minimize that. Could always look at torrenting the disk image if it's too big for a download location.

torpor -- anyway, yeah, so you want to help.. great. Can you build the VM?

If it's not been started by the time I've finished Collector Plus, I'll have a go

pmprog .. I have some experiments, but out of tune a bit right now; teethign babies, then March Break, and a good friend of mine just passed from cancer (too young, mid 30s :/), so I'm rather busy and bummed out at the same time. Thats all excuses, I shoudl get on it, but bummed out all the same Ah, to have the free time I had even 6mo ago

They may be "excuses", but at the end of the day, real life trumps all.

If you're open for testing, maybe I can post something for you in a week or two. (networking, wise.) (Also, need to start considerting overhaiuling the keybindings and layouts .. getting to be a lot of functionality in there, and its not well laid out for it; also, we probably have a better idfea of how thigns are used since it was started so could make it better tuned for those workflows .. at the expense of losing existing skins, mind.)

Yeah, sure, give me a holler when you're ready. I've convinced work to let me try 4 day weeks for 3 months starting next month. That should give me some extra flexibility to work on stuff (and hopefully get Collector out of the way, if I failed to hit the compo date).