Gábor’s Blog

I’ve started to work on moving out the fetch code from bsd.port.mk, which fetches all the distfiles and patchfiles for ports, because this makes bsd.port.mk pretty messy. The fetch code is about 12 pages of A4 and located in bsd.port.mk in bunches, not as a whole. The task is quite simple, I need to convert make if and for cluases to shell, as well as the make substitution expressions.

Besided, I’m still working on DESTDIR. In the exp run, my patches broke some extra ports, that were not broken on the latest i3867 run, so I need to investigate what caused these breakages.

Today (actually yesterday) I sent two patches for our ports infrastructure. One of them is the DESTDIR support. I think everything is done for the infrastructure, but as I said before, some individual ports have to be fixed as well. I did a big effort so that we have to fix the least of them. Erwin took my patches and started to set up an -exp run on the cluster to test if my patch doesn’t break anything if DESTDIR is not set. We will discuss the further things after that.
My another patch makes some standardization and cleanup with ECHO_MSG and ECHO_CMD macros in bsd.port.mk. ECHO_MSG should be used when printing relevant output for the user, while ECHO_CMD should be used mostly internally, e.g. for command-line pipelineing. Accordingly, one should be able to print relevant messages with colors by overriding ECHO_MSG or to write a wrapper for these texts, etc.

Today, I implemented conflict checking in DESTDIR and deinstalling from DESTDIR. Apart from fixing the ports that don’t respect DESTDIR, only one major thing remained, the dependency check. It’s not correct in all aspects. Besides, I’ll try to do some trick with PREFIX as I did with LOCALBASE to avoid modifying a bunch of ports that install files via [pre|do|post]-install target.