| 1 |
Hi Marco, |
| 2 |
|
| 3 |
Marco Clocchiatti <ziapannocchia@×××××.com> writes: |
| 4 |
|
| 5 |
> wonderfull. |
| 6 |
> |
| 7 |
> your eclass is surely better than mine, but I found a little difference. |
| 8 |
> your ebuilds for R packages need a specific SRC_URI, while mine eclass |
| 9 |
> finds automatically the current version of the package by itself. |
| 10 |
|
| 11 |
It looks like your eclass is downloading from CRAN in |
| 12 |
Rutils_src_compile(). For the sake of reproducible builds, ebuilds |
| 13 |
should use local files during the build: |
| 14 |
|
| 15 |
https://devmanual.gentoo.org/ebuild-writing/functions/index.html |
| 16 |
|
| 17 |
Is it possible to instruct R install.package() to "fetch only", or |
| 18 |
"print download URL"? |
| 19 |
|
| 20 |
> I don't know if this is a good or a bad feature, but I think may it be |
| 21 |
> usefull to implement a -9999.ebuild version of the package. |
| 22 |
|
| 23 |
The -9999.ebuild has a special meaning of 'live' ebuild that the source |
| 24 |
is a version-controlled repository. The devmanual discuss the cvs case: |
| 25 |
|
| 26 |
https://devmanual.gentoo.org/ebuild-writing/functions/src_unpack/cvs-sources/index.html#disadvantages-of-cvs-sources |
| 27 |
|
| 28 |
CRAN is not a version-controlled repository. It's packages are |
| 29 |
versioned and released. Therefore I prefer creating versioned ebuilds |
| 30 |
to the live ones. |
| 31 |
|
| 32 |
|
| 33 |
Thanks for your eclass, it is a very neat write-up. |
| 34 |
|
| 35 |
Yours, |
| 36 |
Benda |
Archives