Web of Trust and Freetalk news

Permanent link | Tags: WebOfTrust, Freetalk, Performance, FMS
Creation date: 2020-08-12 19:15:02 UTC

I've written lots of code for Web of Trust and Freetalk since the last update of this flog, but failed to post info about it to the flog because:

  • Posting the news to FMS every week was so convenient :) It's nice to get immediate feedback from the community!
  • The current WoT/FT performance work is very technical and thus posting details of it would perhaps be boring for a broad audience.
  • It's lots of work to do such architectural changes (>1000 hours already) which thus consumes lots of time where there should be a news cycle with lots of news - but all non-technical news you have for the whole time is "I'm making it faster". A difficult task to keep the community informed meanwhile!

But still, not updating the flog in years has grown completely unacceptable to me, I started to feel really guilty about it (sorry!), so I'll just try my best to write a post perhaps every 1-3 months, even if it might be too technical.

I hope that I can find a satisfying balance for my readers. Feedback on FMS is welcome! :)

As a first start, below are links to HTML versions of the current FMS threads to which I've been posting about once per week for the past year.
They can be viewed without FMS:

(Please be aware that I don't know who runs that FMS archive so I can't promise that it's trustworthy! Thanks to them for providing this service though :)

First benchmark of upcoming WoT release; Changelog fixed

Permanent link | Tags: WebOfTrust, Performance
Creation date: 2018-06-28 17:41:21 UTC

After over 1 year of working on the Web of Trust identity downloading code rewrite I've finally reached the point of being able to do a quick & dirty test run as a benchmark.

With 1 hour of uptime of a newly created, empty WoT database I get this:

  • Old code: 132 identity XML files downloaded.
  • New code: 8681 identity XML files downloaded.

So WoT might actually become sufficiently fast to deliver content to new users in a timely manner :)
This will also greatly reduce CPU and network usage:
Instead of subscribing to the USKs of ALL, i.e. well above 10 000, identities, it will only subscribe to identities to which you have assigned a trust. All other identities will be downloaded by USK edition hints obtained from the downloaded trusted identities.

The code is NOT ready for actual usage, it still has dozens of critical FIXMEs and is completely lacking unit tests, but it may not take many more months to finish:
Even though I had been working on it for a year it only took ~30 minutes of bugfixing to get it to actually download stuff - so perhaps my code is less shitty nowadays and it may take less months to test & deploy it :)

Also the build0019 changelog was broken when I published it here some weeks ago, sorry:

The section for issue "[Code quality] Use a human readable file format for serializing Identity XML files to disk (ArneBab, xor)" was randomly cut off and all issues following it were also missing.
This also made the build0018 changelog invisible in the archives.

This has been fixed.

Changelog and important 2018 news

Permanent link | Tags: Ethereum, Altcoins, Freetalk, WebOfTrust, Freenet, Bitcoin
Creation date: 2018-05-29 09:34:50 UTC

As you may have noticed in November 2017 Web of Trust version 0.4.5 aka build0019 was released.
Since I've been focusing on the next release very much and converting the changelog to HTML is quite a bit of work (I'd be very grateful if someone added full Markdown support to FlogHelper!) I had not posted the changelog here yet.
Sorry. In the future I will post the changelogs right when Freenet deploys the WoT releases.

Now even though it has been some months it can be said that the WoT release does provide quite a large amount of interesting new stuff so I would recommend reading the changelog.
It follows below this post.

Beyond that, there are some important news for the future of WoT and Freetalk:
  • Even though there now are enough donations to hire someone and the poll on how to spend them has been finished I won't be an employee of Freenet anymore. It will likely be mrsteveman1.
    I instead will continue my volunteering in a much more independent fashion.
    Both some of the Freenet core developers as well as myself did and do not like the work climate between them and me which has made cooperation beyond deployment of WoT/FT updates impossible at the moment.
    Let's keep the details and blame off this blog for now :) It's enough to say we currently seem socially incompatible.
  • Since I've started measuring the time I contribute as a volunteer I've noticed that I do have a very significant amount of time at hand for volunteering, and I do enjoy it. The number of hours I spent on Freenet in 2017 would amount to 3 months of fulltime work.
    Due to good developments of my personal life I expect to continue being able to donate months of my time per year upon WoT/FT work for the next few years :)
    My goal is to keep shipping 1 Git commit for every day of the year.
As a conclusion I have decided that:
  • I want to integrate more closely with the community.
  • Thus I will update this flog more often.
  • And I will look at the FMS board "freenet" every week from now on.
  • I've resumed Freetalk development to make it even more easy for the community to stick together. It will be a forum system like FMS with the difference of being much more easy to install and use. It'll ship as a regular Freenet plugin and integrate cleanly into its GUI.
  • Having 3 months less time available per year for earning money to pay my bills means that I unfortunately should be begging for donations some more. In the industry the time I contribute would be billed at about $15k to $20k per year, so I hope it's sort of OK to beg.
    Thus besides Bitcoin I have added Ethereum and other altcoins as donation methods to the sidebar at the right. Donations would be very much appreciated :)
By the way: When my computer - a sturdy ten years (!) old ThinkPad T61p - finally died half a year ago I asked for donations on FMS - and some generous anonymous contributor donated enough Bitcoin for a whole new laptop! I'm typing this right on that laptop :) Thank you very very much!

Outlook

I've been working on WoT build0021 for about a year (build0020 will be a minor maintenance build, it's finished and pending testing). It will ship one of the most important big performance fixes which WoT has yet received:
To not subscribe to the USKs of the over 14000 trusted identities anymore - but only to the few identities you have personally trusted, 512 at most.
This work has crossed its peak of complexity, the algorithms are implemented and the most difficult design decisions are done.
From now on it's likely only a few months left of writing glue code, polishing and testing :)

So stay tuned, thanks for your patience and have a nice day :)