Add in-text variables from Code via macros about orgpaper

[tntdynamight]

I found a better solution: With :results output drawer, the extra file isn't necessary, for example in R you can do:

#+BEGIN_SRC R :results output drawer :exports none
# Not APA style, which asks for: between cond1 (M=xx, SD=xx) and cond2 (M=xx, SD=xx); t(df)=xx,p=xx
# One could report (M=xx, CI=xx), or put Means and SD
report_ttest <- function(label,ttest,k=2) {
    decify <- function(x, k) format(round(x, k), nsmall=k)
    prefix <- paste("#+MACRO:",label," ")
    tdf    <- paste("t(",decify(ttest$parameter,2),")=",decify(ttest$statistic,k),sep="")
    p      <- paste(", p=",decify(ttest$p.value,k),sep = "")
    outstr <- paste(prefix,tdf,p,sep="")
    return(outstr)
}
AvsO <- t.test(c(1,3,5),c(6,6,6))
AvsO2 <- t.test(c(1,2,1),c(6,7,6))
AvsOrep <- report_ttest("AvsO",AvsO)
AvsOrep2 <- report_ttest("AvsO2",AvsO2)

print(cat(c(AvsOrep,AvsOrep2),sep="\n"))
#+END_SRC

and You will get:

#+RESULTS:
:RESULTS:
#+MACRO: AvsO  t(2.00)=-2.60, p=0.12
#+MACRO: AvsO2  t(4.00)=-10.61, p=0.00
NULL
:END:

So you can Use (anywhere in Your file):

Apples were no different from oranges in test1 ({{{AvsO}}}), but they were in test 2 ({{{AvsO2}}})

which exports to:

Apples were no different from oranges in test1 (t(2.00)=-2.60, p=0.12), but they were in test 2 (t(4.00)=-10.61, p=0.00)

[vikasrawal]

The org babel docs has this for inline code.

https://org-babel.readthedocs.io/en/latest/eval/#inline-babel-calls

This can be used for in-text variables. What the pros and cons of using this or the macros.

Vikas

[tntdynamight]

The advantage of the macros is that the information flows from potentially larger code blocks towards orgmode. I typically have longer code blocks that I don't want to run all at once or again on export, but I want variables or results to be available in-text and updateable by rerunning the code-block if necessary - but not obligatorily upon export. My examples above are somewhat trivial for brevity, but for longer calculations I haven't found a better way yet to make certain variables from longer code blocks accessible for in-text reference.

[vikasrawal]

I agree, running all the code blocks upon export can soon become problematic. I have never used in-text variables myself, and therefore, speak without experience. Your approach indeed has the advantage that export would be a snap. But my concern is that the document would be cluttered with too many MACROs.

If you use a persistent R session with the document, there is another possible approach. You can create a dataframe with two variables, key and value, in which you store values of each variable that is going to be used in the text. These values could be inserted/replaced by various code blocks. Then, the inline code only needs to read a particular value from a data frame. That should be reasonably quick.

Do you think that can be a reasonable way of doing things?

[tntdynamight]

That should work. I guess it kind of depends where you prefer to have your clutter. My considerations:

1. I haven't actually gotten that to run (you need to run :session and call that session inline). - if you have a working example that would be great.
   1.1. maybe number to string conversion is necessary? -> in that case I prefer to worry about it in the language code block, not in my text.
2. it's still less readable and pretty than macro - compare:
   Number of Subjects was src_R[:session parso ]{as.character(parstab$nsubs)}.
   with
   Number of Subjects was {{{parstab_nsubs}}}
3. Clutter is added by manually defining which variables you want to make available through Macros in the code-block. in the .org file it's not bad at all however, as you store the macro definitions in a drawer anyways.

P.S. The very best way I could imagine would be to allow remote references to table fields from in-text (so far only possible from other tables), ideally being able to call fields by column and/or row names of that table. That way I would include the whole table as supplementary material and easily reference parameters, results, anything from a table, something like: