Karl Millar

[Karl Millar @millar_karl]

IMHO, this is by far the biggest issue facing both the R and python data analysis communities. https://t.co/lNBwi91B9v

[Wes McKinney @wesmckinn]

@basketballvalue @jreback people with skills are the scarcest resource...

[Karl Millar @millar_karl]

.@GenRobertNeller, @USMC, now is your time to show strong and forceful leadership, not issue weak statements. #USMC
https://t.co/B2ZgwCVaSv

[Karl Millar @millar_karl]

Really good, in-depth advice on doing real-world analysis and data science well from some of Google's best.
https://t.co/1szi7MY20l

[Karl Millar @millar_karl]

@bhive01 @hadleywickham @henrikbengtsson Benson Leung has been testing cables so you can be sure you get a good one.
https://t.co/Nm6ct7opXI

[Karl Millar @millar_karl]

rho is the artist^H^H^H^H^H^H project formerly known as cxxr. https://t.co/k41vVSIOrh

#rho #rstats #PrinceGoneTooSoon

[Karl Millar @millar_karl]

It's starting to feel like I'm living in Wellington again...

https://t.co/UmwfBe9s4a

[Karl Millar @millar_karl]

@kennethrose82 Hi Kenneth. I'm fairly sure that Google doesn't have any R users in Aarhus at the moment, sorry.

[Karl Millar @millar_karl]

Note to self: Merge fist. Only then should you do semi-automated refactorings. Too many merge conflicts and manual updates the other way.

[Karl Millar @millar_karl]

Spent all day today writing code that rewrites the source code behind R's builtin functions so that code written in R runs faster.

[Karl Millar @millar_karl]

@hadleywickham Last time I checked (back when SSD was rare), compressing with a snappy-like compression algorithm generally made I/O faster.

[Karl Millar @millar_karl]

@hadleywickham It's fairly data dependent. If your values are integers, you get much better compression helps a lot.

[Karl Millar @millar_karl]

@Collinsjo12 @eddelbuettel @hadleywickham Currently, CXXR performance is worse that R :(

[Karl Millar @millar_karl]

@Collinsjo12 @eddelbuettel @hadleywickham No plans for concurrency at the moment, except for implicit parallelization of vector code.

[Karl Millar @millar_karl]

@Collinsjo12 @eddelbuettel @hadleywickham I'd like to be faster than C++ on well vectorized code.

[Karl Millar @millar_karl]

@Collinsjo12 @eddelbuettel @hadleywickham It's too early to tell. Probably faster that Cpython and slower than WebKit on scalar code.

[Karl Millar @millar_karl]

@eddelbuettel Working on it...

[Karl Millar @millar_karl]

This is amazing. We should constantly test R with it.
http://t.co/VLIfvsSeMd