IRC channel logs

2020-10-23.log

back to list of logs

<zimoun>Hi! In case you missed one output of the recent EOSC Symposium 2020 about “Scholarly Infrastructures for Research Software”: <https://docs.google.com/document/d/1yObRCR7COQctpjMdg-rNenRZ7ZeUZ-u0iyvdpPRK598/edit#> (link accessible until Nov. 10th)
<civodul>zimoun: hey! i tried viewing it but it doesn't work in icecat
<civodul>plus, Google Doc...
<civodul>what's the take-away?
<zimoun>civodul: I received it via mailing list, sent by Roberto Di Cosmo. There is an interesting State of Art review about Archives, Publishers, Aggregators, etc. Well, I can save it, convert to pdf and send you (privately)
<zimoun>civodul: is is possible to send file via IRC (say ERC)?
<civodul>not really
<civodul>maybe you can upload the PDF somewhere?
<zimoun>I have tried to send you via DDC. Is it not working?
<zimoun>EmacsWiki says it is :-) With (require ’erc-dcc)
<civodul>i think DCC is disabled on Freenode due to abuse, isn't it?
<zimoun>ah, maybe… well surely :-)
<zimoun>I will upload it later somewhere and keep in touch this channel
<civodul>looking at https://lwn.net/SubscriberLink/834571/3454cf96085046fc/ on Julia
<civodul>apparently Lee Phillips is the author of several articles in high-profile magazines with titles that make sound Julia awesome (which it surely is)
<civodul>we should learn, we need a writer like this :-)
<zimoun>civodul: yeah Julia is awesome! It’s a Lisp :-) yes, try “julia –lisp“
<zimoun>civodul: thanks for sharing the link. And yeah we should learn from writer like this :-)
<civodul>it's interesting because it's of course a very different project, but it also started in 2012 and it's much more visible than Guix
<civodul>PLs is a domain that's about as crowded/competitive as GNU/Linux distros, too
<civodul> https://en.wikipedia.org/wiki/Julia_(programming_language) even gives download numbers, i wonder how they can come up with download numbers :-)
<zimoun>I do not know if you have seen https://youtu.be/EZD3Scuv02g (sorry for the link, I do not know where it is elsewhere) about Fortress by Guy Steele. Success is sometimes just the good moment.
<zimoun>hum? wikipedia is… wikipedia. And then maybe some marketing team’s stuff… well I wonder too. :-)
<zimoun>and what I retain is that the analogies work well and they are really effective; as Yahourt for bootstrapping, Smoothies for containers, etc. or good artwork as Carl’s video. And in the case of Julia both :-) see https://arstechnica.com/science/2020/10/the-unreasonable-effectiveness-of-the-julia-programming-language/
<civodul>yeah, same author
<civodul>so they have JuliaCon in 2016 (i.e., at age 4) with Steele as a guest
<zimoun>yeah MIT network :-)
<civodul>can we get Steele give a keynote at the Guix Days? :-)
<civodul>or Leroy?
<zimoun>if I remember correctly Sussman was in the committee of Jeff Bezanson PhD (the core dev. of the Julia’s core). Maybe Steve Johnson (FFTW) too. MIT network :-)
<civodul>heh
<zimoun>Leroy? Xavier you mean?
<zimoun>ah and for the fun, the PhD thesis is one issue of Julia https://github.com/julialang/julia/issues/8839
<rekado_>Julia is interesting for bioinfo people because R as a language is so universally disliked
<rekado_>however, my boss sticks with R because of Bioconductor, and he’s pretty … hype-aware and likes to try new things
<rekado_>zimoun: R started out as a Lisp, too!
<rekado_>and Ross Ihaka thinks that a Common Lisp based system will (or should) be the next R.
<rekado_>I think he’s got it backwards, though
<rekado_>he wants a CL thing with R syntax
<rekado_>nobody actually likes R syntax, though
<rekado_>so that’s a bit of an odd goal
<zimoun>rekado_: yeah in the papers about R history are really interesting, especially about link with Lisp.
<zimoun>from my point of view, R does too much thing in my back without raising warnings
<zimoun>and S3 vs S4 is… come on! :-)
*zimoun has to go. Damned parisian curfew!
<civodul>rekado_: yeah R the language is terrible
<civodul>i've seen demo where you could see from a couple of lines the terrible things happening under the hood
<civodul>like variable names automatically used as labels in ggplot
<civodul>i'm skeptical about the "started as a Lisp" meme
<civodul>of course, it's not just about the parens, but still, what is it about? :-)
<rekado_>I take Ross Ihaka by his word: ‘The R engine began life as a very simple Lisp interpreter. The similarities between S and Lisp made it easy to impose an S-like syntax on the interpreter and produce a result which looked very much like S.’
<rekado_>that’s from ‘Back to the Future: Lisp as a Base for a Statistical Computing System’
<zimoun>it is an interesting paper to read by the way
<rekado_>another one here: https://www.stat.auckland.ac.nz/%7Eihaka/downloads/JSM-2010.pdf
<rekado_>‘R did not always look like an alternative implementation of the S language. It started as a small Scheme-like interpreter (loosely based on work by Sam Kamin [4] and David Betz[2]).’
<civodul>ah sure, i trust that it started as a Lisp
<civodul>but then it took a different road
<zimoun>yeah. But then statisticians sent patches… and we all know that stats is complicated
<rekado_>‘The S-like appearance of R was added incrementally. We initially moved to an S-like syntax for our experiments because, although we were both familiar with Lisp syntax, we didn’t feel that it was really suitable for expressing statistical computations. This choice of syntax set us on a path towards compatibility with S. Once initiated, the move towards compatibility with S was irresistible.’
<rekado_>yeah, so I guess the character of that humble origin was short-lived
<rekado_>was it *really* R initially…? Probably not.
<rekado_>it became R only when it wanted to be S.
<zimoun>R is named partly after the first names of the first two R authors and partly as a play on the name of S.[19] – from wikipedia
<rekado_>I feel that Ross Ihaka emphasizes the early Lisp interpreter only to use it as a segue for the argument of using Common Lisp in a future variant of R.
<rekado_>I don’t think that early Lisp interpreter had much of an R flavour (ignoring the syntax, of course)
***card.freenode.net sets mode: +o ChanServ