ratox: 0.2.1 -> 0.4 #32417
ratox: 0.2.1 -> 0.4 #32417
Conversation
| libsodium libmsgpack ncurses | ||
| buildInputs = [ ncurses ]; | ||
|
|
||
| propagatedBuildInputs = [ |
There was a problem hiding this comment.
I believe propagatedBuildInputs only apply to python packages.
There was a problem hiding this comment.
They're used way more in python packaging, while they're normally reserved for ensuring that adding a to inputs also brings in b, as in this case.
There was a problem hiding this comment.
ratox should not explicitly link with libsodium and libmsgpack because it does not include their headers and is not linked with them statically. I have removed the change to libtoxcore.
|
I believe propagatedBuildInputs only apply to python packages.
No, it works for plain mkDerivation too.
You can check it by reverting those lines and compiling ratox.
It will fail.
|
|
How about |
| libsodium libmsgpack ncurses | ||
| buildInputs = [ ncurses ]; | ||
|
|
||
| propagatedBuildInputs = [ |
There was a problem hiding this comment.
ratox should not explicitly link with libsodium and libmsgpack because it does not include their headers and is not linked with them statically. I have removed the change to libtoxcore.
|
`ratox` was able to bootstrap DHT for me.
Maybe some DHT nodes were offline when I tested it. The original patch
is from SLNOS (I prefer things with proper ncurses UI, not with bare
FIFOs), I just saw it pushed to SLNOS and decided to play with it a bit.
If it works for you, you should totally just merge it. I'm too lazy to
play with it again.
`ratox` should not explicitly link with `libsodium` and `libmsgpack`
because it does not include their headers and is not linked with them
statically. I have removed the change to `libtoxcore`.
I see you patched `ratox` for this, which is fine and if it compiles and
runs properly, then LGTM. But I don't see how propagating those things
as in the original patchset could hurt. I could be used to simplify
other expressions using `libtoxcore` that do link those libraries (e.g.
qTox) and then `ratox` doesn't need to be patched (who cares about a
couple of entries in the ELF header).
|
The first 2 or 3 it tried were inaccessible, but it kept retrying and was able to connect.
In general packages should declare all their direct dependencies because it prevents breakage when implicit but necessary dependencies are no longer transitively included. In particular, qtox does |
Motivation for this change
Update.
The problem with both old and new ratox versions is that they have an outdated bootstrap node list and so can't connect to the DHT. If one is to patch the source with an updated node list it works. But I don't want to put a huge patch for that into nixpkgs. Just be aware that it wouldn't strictly "work" as-is if you run the compiled binary. This update is mainly a follow up to #22775 to drop
libtoxcore-oldderivation./cc @peterhoeg
Things done
build-use-sandboxinnix.confon non-NixOS)nix-shell -p nox --run "nox-review wip"./result/bin/)