Proposal - Debian-based automatic build repository

[BlackMarketKarma]

Excellent idea and am willing to help in any way..

[roadscape]

If there is no support for a worker, would the community be OK if I hacked the ppa packages to make myself the referrer on all accounts opened using the ppa packages? I could also do something with openledger where @ccedk pays to maintain and distribute their "official openledger linux packages" which sets them as a referrer, and they promote and distribute those packages through their sites. Then people would see that there are "official" packages out there.

Very good question. If you were to do this I would suggest not using 'bitshares' as the repository name.

Of course, ideally, we'd have a worker-funded repo that's as neutral as possible..

[maqifrnswa]

Sorry if you have to explain it like I'm 5 but would this allow us to plug a raspberry pi into our home network and run a full node on it?  I just listened to the Lets talk bitcoin episode #260 and Dan mentioned running a full node and pointing the light client to it.  This sounds like a more secure way of running bts 2.0 for some people instead of using the web wallets?

Yes. At first just as a package, but if there is interest and funding, as a flashable disk image.  I've made and tested both during the graphene testnets (links to them are in the OP).

Cube has great questions, I collected data over the past year on ppa:showard314/bitshares
https://people.debian.org/~showard/ppastats/
bitshares 1.0 has about 20 consistent users
bitshares 2.0 has 1 or 2 users
additionally, bitshares 2.0 has 1 user using ppa:bitshares/bitshares

the dropoff is probably since the online web wallets are so good, and those that will use the cli tools know how to compile themselves.

I don't know if this accurately reflects the potential, since these numbers are without any advertising or anyone (besides me) telling people about the packages. Since no official site links to them, many don't know they exist or feel as if they are inferior. I agree that the biggest market is windows and apple by far (30x bigger perhaps?), but the early adopters will skew towards linux -- and although users can compile on their own many just want to fire-and-forget, just let the systems upgrade themselves with apt-get as needed.

I can compile libreoffice, but I never have.

I don't want to argue for the worker, I don't need the money -- but if this is something the community feels is worthwhile, great. If not, I can put it on ice for whenever it is needed (if ever). All I want is something we can point to for linux users as "the way" to get binaries "officially" since cryptonomex doesn't have the experience (and it is a distraction to them).

If there is no support for a worker, would the community be OK if I hacked the ppa packages to make myself the referrer on all accounts opened using the ppa packages? I could also do something with openledger where @ccedk pays to maintain and distribute their "official openledger linux packages" which sets them as a referrer, and they promote and distribute those packages through their sites. Then people would see that there are "official" packages out there.

[Brekyrself]

Sorry if you have to explain it like I'm 5 but would this allow us to plug a raspberry pi into our home network and run a full node on it?  I just listened to the Lets talk bitcoin episode #260 and Dan mentioned running a full node and pointing the light client to it.  This sounds like a more secure way of running bts 2.0 for some people instead of using the web wallets?

[cube]

I am using your ppa. Because I have 32bit Ubuntu and old hardware.

We have one user count.  Good. I hope to see more coming out to say they are using the ppa.

BTW, a normal compilation of the source codes can provide for 32-bit or 64bit bts2 program depending on the computer type.  We need to see more people using the prebuilt ppa.

[38PTSWarrior]

I am using your ppa. Because I have 32bit Ubuntu and old hardware.

[kuro112]

I would like to help you make this succeed, im willing to dedicate a small amount of personal time to build you a functional site to make this work and look good, hit me up pal.

[fav]

Could you provide download stats?

[cube]

Given that ethereum is getting a good amount of dev attention, the least we could do is make it easier for devs to try out our platform.

If a 'developer' cannot even do a simple compile,  I fear the kind of codes he/she produces.  I doubt this kind of developers is worth our effort to pursue.  IMHO, this person is more appropriately called a user.

Is there any way to get stats on the number of installs? That would be a great number to have.

Yes, indeed.  I like to support the worker.   But we do need some figures to justify it.

[roadscape]

You maintain the official ppa package for bitcoin? Nice! This would be nice to have IMO.. we could attract countless devs who might otherwise lose interest in trying to get it working. We want people to start playing with it in minutes.. not trying to figure out how to build, registering a forum acct to ask for help, etc.

Given that ethereum is getting a good amount of dev attention, the least we could do is make it easier for devs to try out our platform.

Is there any way to get stats on the number of installs? That would be a great number to have.

[cube]

I think it would be good if we can know an estimate of how many are using a linux pre-made binary as compared to users using self-compiled linux source, windows and mac.  Essentially, we need to know how much demand are there for the services and justify it with the cost to the network (ie inflation).

That is true. Based on the number of people on the forums requesting binaries or help compiling, it does appear to be a barrier to entry. It's unclear the market impact of this barrier to entry. However, the extreme low cost mitigates the risk: the proposal is for 3966 BTS/day which accounts for an annual inflation of 0.057%. In my opinion it is worth less than six hundredths of one percent of inflation to reduce a barrier to adoption.

Most of the linux users I have seen in this forum are technically capable of solving compiling problems with a few pointers given by xeroc or other helpers.  The majority of the users are non-technical and they are using Windows or Mac. 

'low cost' is a relative term which could turn out to be 'high cost' if the benefit (to the network) or demand tends towards zero.  In order to justify any amount spent, the BTS2 shareholders would be asking for a quantification of the benefit(s) to the network ie the revenue or even better, the profit brings to the network should be greater than cost.

[Pheonike]

If constantly maintained, then its worth it. It's one less barrier to entry for people who want to join the community.

[maqifrnswa]

I think it would be good if we can know an estimate of how many are using a linux pre-made binary as compared to users using self-compiled linux source, windows and mac.  Essentially, we need to know how much demand are there for the services and justify it with the cost to the network (ie inflation).

That is true. Based on the number of people on the forums requesting binaries or help compiling, it does appear to be a barrier to entry. It's unclear the market impact of this barrier to entry. However, the extreme low cost mitigates the risk: the proposal is for 3966 BTS/day which accounts for an annual inflation of 0.057%. In my opinion it is worth less than six hundredths of one percent of inflation to reduce a barrier to adoption.

[maqifrnswa]

could you list your monthly expenses?

OP updated with detailed breakdown

Worker proposal costs:
$15/day (~3966 BTS)
New worker proposal every 3 months to adjust for price.
Breakdown:

• Servers: $80 monthly for front-end and back-end servers.
• Domain name: $15 one time
• SSL cert: free (startssl, just to prevent man-in-the-middle attack on the info page)
• Development time: $50/hour based on jobs at freelancer.com
• 10 hours: server set-up development of build scripts (done)
• 1 hour/week maintenance (new versions/bug fixing)
• New distributions and architectures are not included yet. They will add as development time based on actual time spent in future months

Daily expenses over 90 days:
[80*3+15+(10+1*12)*50]/90 = $15.05/day
Rounded down to $15/day

[cube]

I think it would be good if we can know an estimate of how many are using a linux pre-made binary as compared to users using self-compiled linux source, windows and mac.  Essentially, we need to know how much demand are there for the services and justify it with the cost to the network (ie inflation).

[fav]

could you list your monthly expenses?

[maqifrnswa]

Worker created!
OP updated with info.
Thanks, if you have questions please post here

update_worker_votes maqifrnswa {"vote_for":["1.14.6"]} true

xeroc, I'll let you know when it's production ready. It's now fully functional, but still in testing

[xeroc]

Sure .. please ping me once you feel comfortable

[maqifrnswa]

I'll add this to the docs if you don't mind ..

no problem, but maybe wait until it is production ready. I want to make sure there are no obvious bugs first, probably a couple more days. And I'll get a domain name, ssl cert, better hardware once the worker is funded.

[xeroc]

I'll add this to the docs if you don't mind ..

[maqifrnswa]

Worker Proposal for "Official" Debian and Ubuntu Binaries via PPA

To VOTE:

Code: [Select]

update_worker_votes ACCOUNTNAME {"vote_for":["1.14.6"]} true

As of now, no linux binaries of bitshares are available. This will fix that!

This worker will provide repositories for Debian based distributions for BitShares. Debian and Ubuntu are ready on both amd64 and i386. The worker will also implement Mint and extend to ARM processors (Raspberry Pi).

Preview and Installation Instructions:
http://162.243.185.205/
http://162.243.185.205/repos

Raspberry Pi Preview:
https://people.debian.org/~showard/raspbian/


They will be "official" since they are funded by the blockchain with voter approval.

What it does: Pulls from git every hour and checks for a new tag. If a new tag is present, new packages are built. Packages are gpg signed, md5sums are included in release files and signed. The Debian apt tool verifies md5sums and gpg key after download and before installation. Ubuntu packages are hosted and built on Canonical's Launchpad Service (ppa:bitshares/bitshares). Debian packages are built and hosted on the worker's VPS.

Qualifications: maqifrnswa is a Debian developer, has maintained a bitshares PPA for 1.5 years, was a delegate and a witness. maqifrnswa is the founder of the debian bitcoin team and is the maintainer of bitcoin in Debian and Ubuntu.

Internals/Security/Trust: The static pages will be SSL secured reduce chance of prevent man-in-the-middle attack. apt-get transport will follow standard distribution practices (over http). The package list, md5sums, and package info are GPG signed and will be verified automatically by apt's internal mechanisms. Sources are available for inspection via the "apt-get source" command and via the online repository.

Witness: maqifrnswa is currently running a witness. If this proposal is approved, maqifrnswa will serve as witness if needed by the network. If witness numbers should be reduced, maqifrnswa volunteers to be voted out.

Timeline: Prototype site already up and running for testing. Production-level site announced and running for Ubuntu and Debian by December. Mint and Raspbian/ARM by 2016 or early 2016. Future work to investigate: Automatically generate rpm from .deb packages and Ubuntu Snappy packages (container based Ubuntu packages).

Worker proposal costs:
$15/day (~3966 BTS)
New worker proposal every 3 months to adjust for price.
Breakdown:

• Servers: $80 monthly for front-end and back-end servers.
• Domain name: $15 one time
• SSL cert: free (startssl, just to prevent man-in-the-middle attack on the info page)
• Development time: $50/hour based on jobs at freelancer.com
• 10 hours: server set-up development of build scripts (done)
• 1 hour/week maintenance (new versions/bug fixing)
• New distributions and architectures are not included yet. They will add as development time based on actual time spent in future months

Daily expenses over 90 days:
[80*3+15+(10+1*12)*50]/90 = $15.05/day
Rounded down to $15/day

Code: [Select]

create_worker maqifrnswa "2015-10-28T00:00:00" "2016-01-31T00:00:00" 396600000 "Debian/Ubuntu-based PPA" "https://bitsharestalk.org/index.php/topic,19485.msg250031.html" {"type" : "vesting", "pay_vesting_period_days" : 0} true

Code: [Select]

get_object 1.14.6
[{
    "id": "1.14.6",
    "worker_account": "1.2.6004",
    "work_begin_date": "2015-10-28T00:00:00",
    "work_end_date": "2016-01-31T00:00:00",
    "daily_pay": 396600000,
    "worker": [
      1,{
        "balance": "1.13.279"
      }
    ],
    "vote_for": "2:80",
    "vote_against": "2:81",
    "total_votes_for": 0,
    "total_votes_against": 0,
    "name": "Debian/Ubuntu-based PPA",
    "url": "https://bitsharestalk.org/index.php/topic,19485.msg250031.html"
  }
]