Making Reproducible Research Enjoyable | /en/2012/06/enjoyable-reproducible-research/

yihui 2022-12-16 22:27:09

Making Reproducible Research Enjoyable

https://yihui.org/en/2012/06/enjoyable-reproducible-research/

8 Comments

giscus-bot 2022-12-16 22:27:10

Guest *Xiangao* @ 2012-06-26 03:15:00 originally posted:

Bravo~

giscus-bot 2022-12-16 22:27:11

Guest *Sepehr Akhavan* @ 2012-06-26 21:02:29 originally posted:

Perfect...I would suggest you to add this article to the overview of knitr or part of the manual. It gives me the big picture of how to approach the RR concept using knitr and I think it may help others as well...

yihui 2022-12-16 22:27:17

Manuals are boring by definition :) I may consider adding it to the "future knitr book".

Originally posted on 2012-06-27 00:24:12

giscus-bot 2022-12-16 22:27:11

Guest *Ivailo .* @ 2012-06-27 09:57:34 originally posted:

Fantastic! I feel I can now jump bravely into what used to be "the deep water" of RR :)

By the way, is there a way to write Rmd files in languages other than English (i.e. those using Cyrillic characters in particular)? It seems that inserting even a single Cyrillic character into the default *.Rmd file breaks the process somehow, and no output is produced in the preview window of RStudio beyond the point the character was inserted :(

yihui 2022-12-16 22:27:18

Can you take a look at the source code of the corresponding *.html file? Also you may remove.packages(c('parser', 'highlight')) and update.packages(), then try to Knit HTML again.

Originally posted on 2012-06-27 14:15:52

giscus-bot 2022-12-16 22:27:19

Guest *Ivailo .* @ 2012-06-27 15:19:24 originally posted:

I have opened a discussion about the issue at the RStudio support forums (http://is.gd/BznNMo) so I'll post more details there not to clutter your blog with off-topic comments.

giscus-bot 2022-12-16 22:27:12

Guest *Dieter Menne* @ 2012-07-02 09:52:20 originally posted:

I love the work you have done for knitting. I also agree with you that teaching and using latex is not much fun, so it would be nice to have an alternative. However, I do not share your markdown-hype: it may be good for homework and blogs, but try to write a book or a report for the FDA, and you are lost.

Currently missing (approximately in order of relevance)

a) Semi-automatic paging
b) Table facilities with captions
c) Some (latex) float work-alike
d) Cross-references
e) Table of contents, index
f) Justification and micro-justification (e.g. kerning)

So if someone believes that you are close to RR with markdown: don't try it outside of the university habitat.

yihui 2022-12-16 22:27:20

I'm not saying we should replace LaTeX with Markdown. That is impossible. I mean there are more people writing simple documents like homework assignments or short reports than those who write books and FDA reports, and we should not let them suffer for years before they can master LaTeX. In other words, it is not necessary to put an obstacle in front of RR which is just a typesetting challenge and has nothing to do with the spirit of RR.

On the other hand, most of the missing features you mentioned can be found in Pandoc's variation of markdown: http://johnmacfarlane.net/pandoc/ But I just stick with LyX when I need to write LaTeX. I really dislike typing and reading a document full of backslashes.

Originally posted on 2012-07-02 17:18:38

giscus-bot 2022-12-16 22:27:13

Guest *Fabian Greimel* @ 2012-07-17 23:26:47 originally posted:

Hi!
I find your work pretty cool. I am a R and LaTeX beginner and currently working on a bachelor's thesis.
Even though I appreciate that one can create Rnw documents in R-Studio, I am more comfortable in with TeXShop for writing the text sections.

Do you think, you could provide us with a knitr engine for TeXShop? Similar to that with Sweave? (Sweave is not quite suitable for me, because of the lack of the cache function. I didn't get it work)

Thanks

yihui 2022-12-16 22:27:21

You can probably figure it out by yourself. It is not hard. It is nothing but tell the editor to call R via Rscript -e 'library(knitr); knit("your_file.Rnw")'; see http://yihui.name/knitr/demo/editors/

I do not use Mac OS, so I cannot help you here. If you manage to get it to work, please let me know.

Originally posted on 2012-07-18 00:56:12

giscus-bot 2022-12-16 22:27:22

Guest *Fabian* @ 2012-07-28 19:43:05 originally posted:

It is exactly as you said, thanks!

create a file called Knitr.engine in the Engines folder

the content is:

#!/bin/bashexport PATH=$PATH:/usr/texbin:/usr/local/binRscript -e "library(knitr); knit('$1')"latexmk -pdf "${1%.*}"

just had to figure out what to insert in the brackets of knit(...)
now it works fine!

yihui 2022-12-16 22:27:23

Thanks! I have just added it to the knitr website.

Originally posted on 2012-07-28 22:13:04

giscus-bot 2022-12-16 22:27:14

Guest *M. Edward (Ed) Borasky* @ 2012-07-18 00:07:26 originally posted:

RStudio, knitr, Markdown and pandoc are the core of my new project, the Computational Journalism Studio. I'm still hacking / dogfooding on an alpha level, but everything's on Github if anyone wants to join the fun. http://j.mp/CompJournStudio.

giscus-bot 2022-12-16 22:27:15

访客 *志国 马* @ 2012-10-16 11:25:45 写道：

Do you try the org-mode in Emacs? The markup language is pretty simple, and it support different language, such as Matlab, R, Python, C, etc.

yihui 2022-12-16 22:27:24

I have heard about it so many times, and I hope someone can show me directly how it works, or point me to a video, otherwise I have no clue of how to get started. So far it does not interest me. I spent a really long time on getting started with Emacs, and Org-mode is yet anther new world. Now I rarely use Emacs.

Originally posted on 2012-10-19 05:20:40

giscus-bot 2022-12-16 22:27:16

Guest *LeftyGeek* @ 2014-12-05 04:04:09 originally posted:

I'm uncomfortable with the idea of putting all of my (rather large) R code in an RMarkdown document used to generate PDF or HTML. I tried running my R scripts in RMarkdown chunks at the start of my RMarkdown document, with the idea that I would reference the variables created in these scripts to render tables, graphs, and text narration. I as surprised to see that chunks don't reference a common global environment, so my approach doesn't work. My question is: is it possible to separate numerous R scripts, run them, and then access their results in an RMarkdown document referencing variables in a common environment? I think I would prefer to use RMarkdown as a tool to generate the presentation layer of my R programs. Any comments on this approach, and whether it is possible to do this? Thanks.

yihui 2022-12-16 22:27:24

I'm not sure what you tried, but all code chunks do share the same environment by default. R objects created in the previous chunks will be available to latter chunks.

Originally posted on 2014-12-07 05:17:41