Data Science. TileDB. Open Source. Quant Research. R. C++. Debian. Linux. Adjunct Clinical Professor, University of Illinois. Lots of coffee. And some running.
Chicago, IL, USA
Joined March 2007
- Tweets 5,361
- Following 350
- Followers 10,149
- Likes 7,646
Don't do JSON. R's own serialize is much better. You may just need `serialize(x, ascii=TRUE)`. We do this in another context (that is not on CRAN) and it ... just works. Give it a few more whirls, it is worth it.
Friends don't let friends use JSON for large data.
1
Yes, from `qs` and like my `RcppRedis` package this sits on top of base R `(un)serialize` via my `RApiSerialize` package. `qs` is great and clever too, may you can "bend it" for your PostgreSQL use.
1
Not obscure but also ... not obvious. Mapping 'complex' data structures to schemas is hard; running `serialize()` is easier. You can then `rawToChar()` it and should be able to store in a varchar, get it back and call `charToRaw(unserialize()). See my RcppRedis package.
As `nanotime` uses int64_t representation via `bit64::integer64` you can join on that (if you don't convert)!
The `data.table` package does a few things right here (IDate, IDateTime) and has long had both `nanotime` support and joins for time-series.
4
This is #rstats R FAQ 7.31, but in a trenchcoat.
You cannot (generally, reliably) test equality on floats -- which POSIXct is under the hood -- so in turn joins are tricky to impossible. You can truncate down, or do approximate joins, or ...
1
1
5
Thrilled to have learned hat I will get to talk about #r2u at the upcoming Ontario Statistical Software Conference in Toronto! Very much looking forward to it, and now debating whether to be 'in person' or merely virtual... Come join me, and register at the link below. #rstats
A sneak peek at our preliminary speaker lineup for our first annual Statistical Software Conference.
This is a one-day conference bringing together academic and industry participants to share best practices on developing statistical software.
Learn more: bit.ly/3EqA23v
3
12
Also: "Let's just add one more dependency to our #rstats package"
2
3
Rblpapi 0.3.14 on CRAN: Updates and Extensions
Direct and efficient R interface to Bloomberg Terminal
dirk.eddelbuettel.com/blog/2…
#rstats #rcpp
4
7
RVowpalWabbit 0.0.17 on CRAN: Maintenance
R Interface to the 'Vowpal Wabbit' fast out-of-core learner
dirk.eddelbuettel.com/blog/2…
#rstats #rcpp
1
2
Sorry to hear that, but no, not as a general rule (and both `Rcpp` and the packages using it are in generally good shape at CRAN). Can you try smaller and smaller subsets of your package or routines to find a _minimally verifiable complete example_ that differs?
RcppArmadillo 0.11.4.0.1 on CRAN: Updates
R bindings to powerful and expressive C++ matrix library
dirk.eddelbuettel.com/blog/2…
#rcpp #rstats
2
6
There are no short or one-line answers as it depends on the OS, the use case (development too? if so how/which tools?), and more. And #Rstats itself ships with an entire manual devoted to this for a reason:
rstudio.github.io/r-manuals/…
1
2
3
It's a bit of a 'deja vu' -- while I don't have VMs that small (I just use Docker, being on Linux) I answered that same question a few times at StackOverflow etc.
That said, do look into r2u. It's honestly rather good (and I switched all my CI etc to it).
a) That is an out-of-memory error because your VM is too small for C++ compilation
b) solution one: do not compile: sudo apt install r-cran-lme4
c) solution two, much better: look into r2u:
eddelbuettel.github.io/r2u/
#rstats
1
1
10
/cc @EmilyRiederer and 👋 thanks again for talking to my STAT 447 class last fall #rstats
1
1
2
Thrilled to be speaking about #r2u to the #YEGRUG. This should also leave time for demos of the (simple!) setup and its use. I will present remotely (as I have attended remotely) so feel free to join the group. #rstats
More about #r2u as always at eddelbuettel.github.io/r2u/
And I am happy to announce that @eddelbuettel will present r2u at the October #YEGRUG
meetup.com/edmonton-r-user-g…
2
13