<r0ni>
just seems to me that pkgmk -is should do exactly what I ask it, and not refer to pkgmk.conf for directions, as I want it to ignore, else why would i specify it.. just seems odd to me
<jaeger>
agreed, command line args should override the others
<farkuhar>
We could start shipping a pkgmk.conf with more verbose defaults: PKGMK_IGNORE_SIGNATURE="${PKGMK_IGNORE_SIGNATURE:-no}" That way any command-line options are preserved, even if the user uncomments that line in pkgmk.conf. It still doesn't solve the problem of malicious Pkgfiles resetting the IGNORE_* variables, but users should be inspecting the Pkgfiles anyway.
<farkuhar>
Another strategy is the one employed by prt-get: initialize two separate data structures to hold the command-line args and the global config, then consult both of them at the critical junctures in the operation.
<jaeger>
Maybe both
<farkuhar>
TODO38 now has an item based on r0ni's observation, plus some other updates regarding recent discussions.
groovy2shoes has quit [Remote host closed the connection]
groovy2shoes has joined #crux-devel
<jaeger>
suitesparse is failing to find lapack even with lapack installed, looks like
<beerman>
i'll check
<jaeger>
I saw the same in a fresh container, though I haven't had time to dig deeper yet
<beerman>
I can confirm
<beerman>
I feel like the easiest fix is to temporarily downgrade lapack.. the release looks weird if you look closer. there is this https://github.com/Reference-LAPACK/lapack/pull/1095 which is not included in the release from two weeks ago but it doesn't help anyway, I am too tired for this tonight
<jaeger>
sorry, heh
<jaeger>
I'll downgrade it for now when I'm on that system again
<jaeger>
Ah, never mind, I see you reverted it already