Open mcfrank opened 9 years ago
I think the solution is to make this into an R package, and I've mostly converted it to that form to try out package creation. Using it as a package in various projects is much saner/cleaner than pulling in a script -- it can just live on GitHub and people can use devtools to install it. The package structure and checks also have very good side effects in making the code better and better-documented -- writing out all of the documentation made me notice and fix a bunch of things that were sloppy, ugly, or buggy (for example, the function for computing standard error of the mean had a bug that made its result straight up wrong if there were any NAs in the data...).
The main drawback of this approach is that creating a package of "various things our lab uses" is sort of weird, but it's not like we have submit it CRAN or anything, we can just use it internally and it'll make our code better.
sounds like a great solution!
Mike
On Wed, Jun 17, 2015 at 11:55 PM Mika Braginsky notifications@github.com wrote:
I think the solution is to make this into an R package, and I've mostly converted it to that form to try out package creation. Using it as a package in various projects is much saner/cleaner than pulling in a script -- it can just live on GitHub and people can use devtools to install it. The package structure and checks also have very good side effects in making the code better and better-documented -- writing out all of the documentation made me notice and fix a bunch of things that were sloppy, ugly, or buggy (for example, the function for computing standard error of the mean had a bug that made its result straight up wrong if there were any NAs in the data...).
The main drawback of this approach is that creating a package of "various things our lab uses" is sort of weird, but it's not like we have submit it CRAN or anything, we can just use it internally and it'll make our code better.
— Reply to this email directly or view it on GitHub https://github.com/langcog/Ranalysis/issues/4#issuecomment-113027845.
Alright, I pushed the package I made to a new repo https://github.com/langcog/langcog
It has several different components:
There are a few function that were Ranalysis that I didn't include because they seemed questionable to me, but I'm happy to discuss if/how to include them if someone uses / would use them.
Awesome. So we use devtools to install it?
Mike
On Jun 18, 2015, at 12:16 AM, Mika Braginsky notifications@github.com wrote:
Alright, I pushed the package I made to a new repo https://github.com/langcog/langcog
It has several different components:
multiboot: non-parametric bootstrap with multiple sample statistics (from Ranalysis) util: miscellaneous utility functions (from Ranalysis) solarized_palette: colour schemes based on the solarized accent colours (from wordbank) clm: linear models with coefficient constraints (from wordbank) There are a few function that were Ranalysis that I didn't include because they seemed questionable to me, but I'm happy to discuss if/how to include them if someone uses / would use them.
— Reply to this email directly or view it on GitHub.
Yup, like its readme says, you can just run
install.packages("devtools")
devtools::install_github("langcog/langcog")
It might be good to do a bit of code review or something before advertising the switch?
Sorry,I wanna to convert an idf file to plain text and have no access to idf converter...does your code do this...please help me how?I tried this in R but unfortunately no success...