PS 1Gotta admit that the point about the object systems in R is somewhat spot on
PS 2 #php gets a dishonorable mention
PS3 I will continue to find ways to continue using all 4 of the aforementioned languages, as they are all performant and deliver in complementary ways.
I think it was a deliberate choice to make it easier for people to transition to Perl.
People were using sed, awk, and other tools until Perl came along and could do many jobs all by itself.
I think Randall Schwartz got in trouble on Usenet for popping in with a Perl solution to problems when someone was expecting sed, awk, or something in shell instead.
@mjgardner@Edwardsmoon@BobOHara@bduncan It is a step backwards, likely explained by people migrating to #bioconductor (in #rstats) or #biopython and wanting to break from clean from #perl. However neither R nor python provided the facilities that perl has to deal with features in text and flat text files.
@mjgardner@Edwardsmoon@BobOHara I guess so. One of the more technical reasons is that most applications in this domain are built as Unix filters that communicate to each other via stdout, so one liners with sed/awk provide an easy way to do throwaway programming. Encouraged by the need to innovate (for fame and grants),ppl are then forced to extremely heavy DSLs and frameworks to control dataflows. The writing for such approaches is already on the wall IMHO (but I will stop rambling nos)
@mjgardner@ChristosArgyrop@Edwardsmoon@BobOHara@bduncan they’re just following the natural progression from shell to awk, with Perl being the inevitable next iteration. Biologists do like to say that ontogeny recapitulates phylogeny
@mjgardner@ChristosArgyrop@BobOHara@bduncan
Isn't there a joke in the documentation or source code somewhere, I think about how $/ can't be a regex, where Larry says "I had to let sed still be better at SOMETHING!"
@ChristosArgyrop We got it. You hate #Rstats. However, your opinion is based on a poor knowledge of the ecosystem. But more important, you were irrespective to the thousands of @rstats developers to devote their time to open source software.
@ChristosArgyrop@rstats In that case, I would recommend using {renv} to create virtual environments, and {box} for importing modules and packages preserving namespace. Also {targets} for your pipelines. With this you will overpass some of the shortcomings you mentioned in your post.
@jospueyo@ChristosArgyrop@rstats I think you should read Christos toot again. The blog post he linked was written by someone else. Christos seemed to be amused that he found an article that talks thrash on two languages he uses...
@jospueyo@ChristosArgyrop@rstats I respectfully disagree. I like R and use it on an everyday basis, but most of his points are rather reasonable. It is important for a community to be able to handle constructive criticisms, and to understand that decisions like "let's not fix interface bugs" can be detrimental 20 years later.
R has many flaws, but the one it doesn't have is to pretend it's a one-stop-fix-all powerhouse. It's a tool for reproducible and interactive advanced data analysis, and for that, it's still unbeatable.
@ChristosArgyrop I bought the R Package by Hadley Wickham for reading on an airplane. I tossed the book in the garbage at the end of the journey after uninstalling R.
@Mehrad@Osunderdog@ChristosArgyrop locally could be less than 5min. You only need a DESCRIPTION file and devtools package. I create internal private packages for each analysis project so I can autoload of functions in R folder, run unit tests and create documentation from comments.
Add comment