Lluís Revilla

Bioinformatician 📢 about OSS and #rstats, some analysis 📊 & random thoughts and ideas 🎲
My opinions/comments not those of my employer (Roche)

Website: https://llrs.dev
Github: https://github.com/llrs

Public Key

npub1hzntl2k9gav5ryhq3sx6ecfs9zuduglqfqgr6467ksrqece8skuqq8f23r

NIP-05 Address

lluis_revilla_at_fosstodon.org@momostr.pink

Profile Code

nprofile1qqst3f4l4tz5wk2pjtsgcrdvuycz3wx7y0sysypa2a0tgpsvuvnctwqpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0ds7n4nxj

Publishing to

nostr-pub.wellorder.net

nos.lol

relay.damus.io

Author Public Key

npub1hzntl2k9gav5ryhq3sx6ecfs9zuduglqfqgr6467ksrqece8skuqq8f23r

Show more details

Last Notes

2026-06-02 12:58:35 UTC

- reply

not sure I understood all the requirements, but with eval()?

2026-05-20 07:23:55 UTC

- reply

I am not that familiar with the internals but given that it implements a new engine to save plots I am sure there is something to pass the plots to the device/engine. I thought it would be helpful... 🤷‍♂️
Good luck, I hope you find a way to solve this

2026-05-19 21:44:23 UTC

- reply

Checking how ragg does it might be helpful.

2026-04-10 13:37:34 UTC

- reply

I think it depends on the server used. I saw some toots using it.

2025-09-11 08:07:29 UTC

- reply

I saw this recently from @npub157d…78z3 : https://mastodon.social/@coatless/115178430811415118

2025-07-31 09:50:25 UTC

- reply

I heard arguments in favor of this rounding method because it is more precise as it will be up or down in similar proportion in data like yours, so further analysis will not deviate much more from the original values

2025-07-07 16:32:02 UTC

We are looking for some feedback from new(ish) #rstats package developers for a CRAN task view: https://github.com/hturner/pkg-dev-ctv/blob/main/proposal.md
We would appreciate you could check it and let us know if something is unclear or as a new R package developer you miss something.

2025-06-08 15:04:39 UTC

- reply

I found it recently but didn't have time to compare it against other licenses or check the content. I guess it could be used for those software written with EU grants.

2025-04-28 22:14:47 UTC

Now that Spain is on the #PowerOutage news, I wrote a small #rstats package to retrieve some data from REE: https://ropenspain.github.io/resios/
I hope (data) newspapers like @civio will make good use of it. I get daily estimation of prices with it, so it should work for indicators 545 and 544. Let me know if there is something that doesn't work.

2025-03-14 10:56:52 UTC

- reply

Or you don't feel like going to USA you might prefer to come to #Barcelona on September for #rstats and #bioinformatics at #EuroBioc2025: also with time for abstract submissions https://eurobioc2025.bioconductor.org/
The weather is still sunny and hot on September 😉

2025-03-13 22:34:04 UTC

#rstats Bioconductor conference is with Galaxy #bioinformatics. There is still some time for early registration (and abstract submission). https://gbcc2025.bioconductor.org/

2025-02-13 06:14:27 UTC

- reply

I am part of a the Bioconductor community board. I would like to hear more on how we can improve Bioconductor. While I have several ideas hearing from other (dissatisfied) useRs would help a lot. Please send me an email, with your thoughts (and if you wish more details about this case or others)

2025-01-26 16:22:27 UTC

- reply

I'm not convinced these reasons are enough:
1. A book can have a DOI and ISBN. Bibliography and metainformation can be stored/shared in ORCID and other places.
2. It can be stored on Zenodo and other official/state libraries which is as much guarantee as there is.it could be served online as its done for other books.
3. A book can be cited and it can be reviewed on certain publishers. Or by the community itself.
But I don't know much about the publishers.
@npub1r5r…u33l

2025-01-25 14:06:58 UTC

- reply

My (current) idea is not to make a long book or publishing a journal article with it review process and format limits. What do you like from a journal article for this?
I would collect the view of several people and group by theme or dedicate one chapter per person. But not order chronologically all the events.

2025-01-25 13:15:30 UTC

Reading about the Fritz Leisch, I wonder how many people were important to shape #RStats. How many histories about their contributions to R and the large scientific or computational areas we are missing or the influence on other people.
I am considering writing a book about those core people (before it is too late) and their profesional or personal relationship around R. At the moment, I have 8 questions I'd like them to answer. Would people be interested in a book like that?

2025-01-23 23:17:01 UTC

- reply

I'm not sure if this is an R design. In principle ?warningCondition has the same arguments as ?errorCondition so I think they could do the same.
This reminds me that people coming from other languages usually don't like that most exception handling is mostly based on strings and not classes. But base R has had some commits adding classes some errors. Perhaps this will come to warnings too or other "meta" information...

2024-09-21 11:29:49 UTC

- reply

nice, didn't know about base::.packages()

2024-09-21 08:54:17 UTC

- reply

never thought about that. Is it particularly of .First on .Rprofile or just .First? I think it is related to .First being executed when some package environments aren't loaded (just base if memory serves me well).
There was that diagram of what happens at R start up by @npub1dal…7r8z that might help too (I don't remember where it was published)

2024-09-16 15:08:46 UTC

The results of the #rstats #CRAN survey is available 😃 https://github.com/r-devel/cran-cookbook/wiki/CRAN-Cookbook-Survey-Results
I thought more than 126 people would have answered. It seems that working on the R package development task view would be appreciated. But there are a lot of suggestions for the CRAN team too.
Many thanks @npub1epc…9tam and team for sharing it.

2024-08-06 20:24:20 UTC

- reply

Kurt mentioned that instead of using some machine learning to annotate/cluster package he sees the CRAN Task Views (CTV) as a better approach. He mentioned that only around 20% of packages are in a CTV. So expanding the coverage of CTV would help users (see slide 38 from: https://static.sched.com/hosted_files/userconf2024/41/user2024_CRAN.pdf).
Currently there are 44 CTV, but you can propose new, and see proposals/drafts via https://github.com/cran-task-views/ctv

2024-08-06 20:19:03 UTC

- reply

Yes, users have requested to have HTML manuals and CRAN wants that too. R now has pkg2HTML to serve them. But CRAN wants to link the documentation well. Among other issues (accessibility, links to Bioc..) there are ~300 packages that mention some code reference but not from which specific package. The idea is to document that so the links can be resolved to a single package.one example: instead of a link to %>% document your link as magrittr::`%>%` (as it is reexported in other packages)

2024-08-06 17:36:59 UTC

You can improve CRAN with these 9 tricks 😉
As it has been discussed these days at #rstats there is much to improve #CRAN, both technically and non-technically. Kurt Hornik, one of the founders of CRAN, listed in his last slide of useR!2024 what *we* can do
My personal recommendation is to start small and with what you have at hand (resources available: time, interest, experience, ...). The hardest part in open source is not naming, but collaborating and long term maintenance
https://cdn.fosstodon.org/media_attachments/files/112/916/251/698/001/746/original/3476e1b5abb9029f.png