Re: Bioconductor API Bump to 3.13
On 01.09.21 17:47, Nilesh Patra wrote:
On 1 September 2021 6:00:44 pm IST, "Steffen Möller" <steffen_moeller@gmx.de> wrote:
which routine-update could then iterate over?!?
Uhhmmmm, no. Routine-update is a nice helper, but its in now way fit to
to do simply
for pkg in PACKAGES ; do
routine-update
dput
Its not made for this and you definitely should inspect the result of
every single run.
I was more thinking of a
for pkg in PACKAGES ; do
if ! routine-update; do
echo "E: First package to fail is $pkg"
break
fi
done
for pkg in PACKAGES ; do
if happy(pkg); then dput; else <manual rework>; dput; fi
done
This is definitely not good.
This doesn't inspect autopkgtests,
The inspection would be part of happy(pkg)
doesn't let you do any QA work on it, it wouldn't check if some new file has a different license which needs to be accounted in d/copyright.
It doesn't account for the fact if some dependency needs to be rebuilt/updated first with 3.13 API.
Directly dput'ing is not good, I can 100% concur with this.
Nobody was suggesting unseen dputs. My question for you was if you would
appreciate an effort to autobuild as much as possible and adjust patches
where builds fail.
I just had a look at the top of the list, i.e. r-bioc-biocparallel,
which leaves the source directory changed upon d/rules clean by removing
the auto-created configure file and leaving config.log in. I have now
created a +ds version of it for that reason, but am not sure if that is
not possibly more than we want to do.
Best,
Steffen
Reply to: