Profile cover photo
Profile photo
Tobias Oetiker (Tobi)
in pursuit of happiness
in pursuit of happiness
About
Posts

Post has attachment
A day on the Albis
Albis 2016.05
Albis 2016.05
photos.google.com
Add a comment...

I finance a large part of my work on RRDtool and my other
OpenSource project through sponsorship. The income from the OSS sponsorships contributes to the maintenance and long term support of my OpenSource projects as well as to the creation of new tools.

It would be great to see join in for the 2016 season.

Offers remain the same as in previous years. I am looking for individuals and organizations taking up a one year sponsorship for RRDtool, SmokePing and MRTG. For each of the projects there are:

* 1'000 USD Silver Sponsorships (you get mentioned on the respective sponsor page on the website, with logo and link to your homepage).

* 10'000 USD Gold Sponsorships (you get mentioned on all pages of the project website, with logo and link to your homepage). Gold Sponsorships are limited to two per project.

Anonymous sponsorships or other arrangements can be defined upon request.


** A look back on my 2015 OSS activities **

* In spring 2015 I finally released RRDtool 1.5.0, quickly succeeded by a bunch of point releases, taking care of some regressions. With version 1.5 there is now a test-suite which enabled us to catch a bunch of portability issues that were fixed in point releases.

RRDtool 1.5 now runs not only on Intel Linux but also Sparc Solaris, the Hurd and everything inbetween.

* As an offshoot from a commercial development project I have published RrdGraphJS, a library that makes it very easy to get interactive RRDtool charts. The only thing that is required is a serverside component that generates RRDgraph charts on demand:

https://github.com/oetiker/RrdGraphJS

I also produced a little video about this new library to show off its capabilities:

https://www.youtube.com/watch?v=3UgDXKskZME

* Along side the 1.5 stability work 1.6 has come a long way. The main feature of 1.6 will be that it is fully thread-safe out of the box for all functions. A bunch of other new features will round out the release. I expect to publish this early in 2016.

* The the future direction of RRDtool development has not progressed as expected, but I am in the process of re-focussing this effort. The objectives are still the same: more flexible data storage and higher performance. But in contrast to the earlier approach of fully rewriting RRDtool from scratch, the current idea is to follow a starvation approach where new functionality shows up in regular releases, improving in reach and stability until it is 'starving' the existing function and will take over its place.

The advantage of this approach is that there will not be a disconnected development branch, but rather experimental features showing up in regular releases.

* I have also published or substantially updated a number of new tools on github:

LdapPasswdWeb - an online ldap password changer with the ability to change samba passwords too.

https://github.com/oetiker/LdapPasswdWeb

callbackery - a toolkit for writing modern web apps in Perl without touching any html/css or even JavaScript for that matter.

https://github.com/oetiker/callbackery

GpfMeter - Grounded Performance Feedback Meter - A windows desktop problem reporting tool for getting a handle on unspecific performance problems ...

https://github.com/oetiker/GpfMeter

All my current OSS coding projects can be found on

https://github.com/oetiker

I wish you all good success in 2016
and all the best

Tobi Oetiker
Add a comment...

Written any cool new OSS tool this year? Present it @ The #LISA15 GitHub Showcase BoF https://www.usenix.org/conference/lisa15/birds-feather-sessions#github
Add a comment...

Post has attachment
Photo
Add a comment...

these are almost equivalent ...

ruby - def x(y=[]) p y << 1; end
perl6 - sub x ($y = []){ $y.push(1).say }
python - def x(y=[]): y.append(1);print y

x();x();x()

except, that python is right ... or so they claim ... and I can see why they think so, but it's pretty inconvenient, and I just spent 4 days learning the finer points of python until I spotted the problem ... grrrrrrrr
Add a comment...

Post has attachment
wählen gehen!
Add a comment...

Post has attachment
three days hiking in the greina
PhotoPhotoPhotoPhotoPhoto
9/29/15
42 Photos - View album
Add a comment...

Post has attachment
PhotoPhotoPhotoPhotoPhoto
28. Oltner 2-Stunden Lauf 2015
280 Photos - View album
Add a comment...

Post has attachment
PhotoPhotoPhotoPhotoPhoto
The Sad Caterpillar
6 Photos - View album
Add a comment...

Post has attachment
If we're stealing merely as part of kmem reaping (versus stealing to assure forward progress), we don't want to actually steal any active htables.  (Stealing active htables merely to give memory back to the system can inadvertently kick off an htable crime wave as active processes repeatedly steal htables from one another, plummeting the system into a kind of HAT lawlessness that can become so violent as to impede the one thing that can end it:  the freeing of memory via ARC reclaim and other means.)  So if we're reaping, we limit ourselves to the first pass that steals cached htables that aren't in use -- which gives memory back, but averts the entire breakdown of social order. -- Brian Cantrill
Add a comment...
Wait while more posts are being loaded