CRAN vs GitHub? about ctv-archaeology

[nfrerebeau]

An additional check could be to ensure that GitHub projects are at least correctly structured and can be installed as a package.

[SCSchmidt]

I would not separate them, but just re-ordering to put the CRAN packages first is fine with me.

[benmarwick]

Yes, good idea, thanks for the suggestion, this seems fine to me too, to order the packages within each section to put the CRAN packages first. Another criteria for ordering could be use metrics. Both @sebastien-plutniak and I have done some work on this, to identify what packages are most often used. Many of the github packages on our list have not been cited at all (and some are inactive, as you say), so probably the ranking would be the same as a CRAN-first ranking. I think going with CRAN packages first is fine for now.

[SCSchmidt]

I've done the re-ordering of the packages here, and put CRAN first: #65

[nfrerebeau]

I retrieved the date of the last commit of each GitHub project. Only 1/3 of the projects have had any activity in the past year (although this isn't a perfect indicator, some stable projects don't necessarily need to be very active).

repo	last_commit
geanes/bioanth	2015-02-27
ISAAKiel/magAAR	2017-06-10
mrecos/signboardr	2017-11-04
sfsheath/cawd	2017-12-21
wccarleton/lamap	2018-06-27
ercrema/HERAChp.KandlerCrema	2018-09-20
Johanna-Mestorf-Academy/sdsanalysis	2018-12-13
nevrome/varnastats	2019-01-18
maciejkasinski/quantatools	2019-02-17
eScienceCenter/SiteExploitationTerritories	2019-04-09
cornelmpop/Lithics3D	2019-04-21
ISAAKiel/lecAAR	2019-07-29
benmarwick/signatselect	2019-12-03
SCSchmidt/percopackage	2020-04-28
ercrema/cTransmission	2020-05-14
Andros-Spica/cerUB	2020-06-06
ISAAKiel/shapAAR	2020-07-08
ISAAKiel/quantAAR	2020-07-09
lsteinmann/clayringsmiletus	2020-11-24
joeroe/swapdata	2021-03-25
davidcorton/archSeries	2021-04-28
DCPollard94/knossoscemeteries	2021-06-22
ISAAKiel/pathAAR	2021-09-03
nevrome/bleiglas	2021-09-15
yesdavid/outlineR	2021-11-14
ISAAKiel/aoristAAR	2021-11-30
mrecos/klrfome	2022-03-28
benmarwick/mjbnaturepaper	2022-05-02
benmarwick/evoarchdata	2022-06-16
people3k/p3k14c	2022-09-21
tonydoss/UThwigl	2022-11-02
xronos-ch/xronos.R	2022-12-21
lsteinmann/datplot	2023-02-18
benmarwick/roev	2023-02-20
joeroe/c14	2023-05-12
joeroe/stratigraphr	2023-05-12
joeroe/islay	2023-05-22
ArchaeoStat/ArchaeoData	2023-06-02
mauricio-camargo/rysgran	2023-06-17
joeroe/rintchron	2023-07-26
ropensci/c14bazAAR	2023-09-05
UCL/ADMUR	2023-09-11
joeroe/fieldwalkr	2023-10-11

[SCSchmidt]

So, do you suggest we use this as a basis to re-order the list? I was thinking about the order of packages yesterday as well and wasn't sure we had agreed on anything yet. So far I think we said 1. CRAN, 2. core, 3. "rest", and "rest" is an awful lot. It would be easy to use date of last commit, and update this once a year for the ones that are active? 🤷 It's also a way to not use order as an indicator of quality or endorsement.

[nfrerebeau]

To be honest, I'm not sure exactly how to use this information 😅 I looked at the last commit dates out of curiosity, thinking it might be a way to see what projects might have been abandoned/retired. It's a good point not to use order as an indicator of quality or endorsement (another possibility would be to sort randomly?).

[benmarwick]

Yes, this is an interesting question. I agree with starting the order with CRAN and core packages, and for the rest I think no sorting is fine. I took a quick look at some other CTVs, and I see alphabetical order in some, and no obvious order in others. My impression is that most users are not expecting any meaning in the order of packages after the first 1-2 in the list. Ours is very short compared to most CTVs, so a user can probably read the short descriptions for each package and make a choice based on that, rather than relying heavily on our ordering. We could put a sentence in the preamble of our CTV telling users that packages are ordered first by those on CRAN and that we consider core, and then randomly (or alphabetically). Then the user wont mistakenly read any meaning into the order that they see.