Open MSoegtropIMC opened 2 years ago
Your log doesn’t include the opam commands being executed. Could you send a link to this ./coq_platform_make.sh
script?
The script is rather lengthy (~ 20 files). The call which is failing is here:
https://github.com/coq/platform/blob/18992d85bc2df3345fb0d71c2fcf5a4c9d15b272/shell_scripts/build.sh#L35 - a plan opam install <some packages
.
The variable "${PACKAGES//PIN.}" is defined here: https://github.com/coq/platform/blob/995ccb5563e820f6fd2fa84f4c068a6eda2e2a08/package_picks/package-pick-8.15%7Epreview1.sh
Note that the PIN stuff is ignored in a parallel install.
Some code forensics:
because the packages are explicitly being installed, opam doesn't display the reason for recompilation
Is there a way to show the reasons anyway? Maybe with -v?
I have an odd issue: I have scripts which call opam install with a largish number of packages (~50) to leverage parallel install. Sometimes opam then wants to recompile a good fraction of the packages, even though they are already installed and to my best knowledge nothing changed (I do run opam update before). The very odd thing is that when I abort the opam install with Ctrl+C and try the same command again, it doesn't try to reinstall and just says that all packages are already installed. I appended a log.
superflous_update.log
I have the impression that this started after I changed my scripts to switch to a certain switch with
rather than
which I did because I sometimes run several opam installs on different switches in different shells. Should I maybe do both?