[Rd] Proposed diff.character() method

[Simon Urbanek]

Arni,

I appreciate your idea, but I would argue that you are really writing a new function that has nothing to do with the diff() function in R. diff() computes (variably lagged) differences between elements of a vector, so if you were to even contemplate diff.character, it would certainly have nothing to do with files (since character vectors are not files in the first place).

Therefore I think it's a great idea, but you probably want to start with a function that compares character vectors element by element compare(x, y) and returns something suitable and the write something like file.compare <- function(a, b) compare(readLines(a), readLines(b)). This has nothing to do with the diff() function R, but could be a nice package. Or, you can have a look at diffobj::diffFile().

Cheers,
Simon


> On Mar 9, 2022, at 5:24 AM, Arni Magnusson <thisisarni using gmail.com> wrote:
> 
> Dear R developers,
> 
> Recently, I was busy comparing different versions of several packages.
> Tired of going back and forth between R and diff, I created a simple
> file comparison function in R that I found quite useful. For an
> efficient and familiar interface I called it diff.character() and ran
> things like:
> 
>  diff("old/R/foo.R", "new/R/foo.R")
> 
> Before long, I found the need for a directory-wide comparison and
> added support for:
> 
>  diff("old/R", "new/R")
> 
> I have now revisited and fine-polished this function to a point where
> I'd like to humbly suggest that diff.character() could be incorporated
> into the base package. See attached files and patch based on the
> current SVN trunk. It can be tested quickly by sourcing diff.R, or by
> building R.
> 
> The examples in diff.character.html are somewhat contrived, in the
> absence of good example files to compare. You will probably have
> better example files to compare from your own work.
> 
> Clearly, the functionality differs considerably from the default
> diff() method that operates on a single x vector, but in the broad
> sense, they're both about showing differences. For most programmers,
> calling diff() on two files or directories is already a part of muscle
> memory, both intuitive and efficient.
> 
> There are a couple of CRAN packages (diffobj, diffR) that can compare
> files but not directories. They have package dependencies and return
> objects that are more complex (S4, HTML) than the plain list returned
> by diff.character().
> 
> This basic utility does by no means compete with Meld, Kompare, Emacs
> ediff, or other feature-rich diff applications, and using setdiff() as
> a basis for file comparison can be a somewhat simplistic approach.
> Nevertheless, I think many users may find this a handy tool to quickly
> compare scripts and data files. The method could be implemented
> differently, with fewer or more features, and I'm happy to amend
> according to the R Core Team decision.
> 
> In the past, I have proposed additions to core R, some rejected and
> others accepted. This proposal fits a useful tool in a currently
> vacant diff.character() method at a low cost, using relatively few
> lines of base function calls and no compiled code. Its acceptance will
> probably depend on whether members of the R Core Team and/or CRAN Team
> might see it as a useful addition to their toolkit for interactive and
> scripted workflows, including R and CRAN maintenance.
> 
> All the best,
> Arni
> <diff.character.txt><diff.character.patch>______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel