If I recall correctly the NVidia driver requires it too. I had some other issues with other ports/packages too, but I can't remember. In any case, GENERIC has COMPAT_FREEBSD4 through COMPAT_FREEBSD13 included, so you're not going to run into that issue.
I have few large ports to recompile still, though there was only one port that didn't reinstall cleanly and will probably end up being a bug report. Looks good so far.
make delete-old-libs. Not much of a difference if one is already building from src though.
pkg upgrade -f following the handbook.ports-mgmt/portsnap is back.
Oh thanks, I should probably look into doing so as I am running 14.0 for the Graphics drivers!
So portsnap was going to be removed from base because of move to git.
https://git-scm.com/book/en/v2 is the freely available through Creative Commons book on Git. Even if all you need is the first 3 chapters for most purposes, it's not easy. Simply downloading ports tree for that task, isn't so bad. The rest of the tasks, you can't even find what you want even when looking at the desired commands.
--depth 1 and --single-branch when doing the initial clone, because you likely don't want to download the entire history of the portstree to your local system. # /usr/bin/time git clone --depth 1 --single-branch https://git.FreeBSD.org/ports.git testing
Cloning into 'testing'...
remote: Enumerating objects: 192788, done.
remote: Counting objects: 100% (192788/192788), done.
remote: Compressing objects: 100% (180691/180691), done.
remote: Total 192788 (delta 11274), reused 117507 (delta 5634), pack-reused 0
Receiving objects: 100% (192788/192788), 83.07 MiB | 5.84 MiB/s, done.
Resolving deltas: 100% (11274/11274), done.
Updating files: 100% (156764/156764), done.
26.31 real 7.35 user 4.71 sys
What does Github have to do with this?
git clone -b main --depth 1? I've read somewhere that if you would include portsnap database disk usage to the downloaded ports tree, total disk space allocation is similar to the git repo. --depth 1 --single-branch is essentially what poudriere ports is doing when fetching a ports tree as I learned a while ago [1] git -C /usr/ports pull seems faster than portsnap auto. make -C /usr/ports fetchindex is required, though.
Or
make index If you have set different defaults using DEFAULT_VERSIONS you really need to build INDEX yourself.Really? That doesn't sound normal. For the test, I just updated my ports tree, which was last updated about one month ago.
git pull took exactly 27 seconds to complete... ok, this computer uses quarterly ports, so there's not that much to update, but also on other machines tracking latest ports it never took more than a few minutes for me, even for the initial checkout.I don't think so. Portsnap can't deal with quarterly branches, which every now and then is a source of avoidable problems for users, who often aren't aware that portsnap downloads latest ports while pkg tracks quartely by default. Git is just a better default option, and is not complicated to use either (new users will just follow the relevant handbook section).
Nothing with portsnap but the fact that I can't seem to consistently view files there makes me mad.
"Need" INDEX? probably not. For me it's always been a handy way to quickly look at a ports dependencies and such because I simply like using grep.
Need? No.
That's the only reason why I generate it after updating the ports tree. Works fine without an index, it's just terribly slow compared to looking up with an INDEX.
INDEX is useful for
psearch. That's a good way to find programs, and it lets you do a search of the longer description as well, though it's better to limit those searches to relevant categories. That's easier to use than FreshPorts and better than pkg search.The only setback of Portsnap. I don't see why there can't be a way to make the branch a choice in that utility.