Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
to promote the connection to HPN.
|
|
|
|
|
|
|
|
|
|
|
|
Bump to 7.9p1 and openssl 1.1.
|
|
|
|
|
|
I am not happy with the solution though as I haven't found a clean way
of forcing openssh to build against a specific openssl version.
I tried pkg-config, but either it doesn't work or I am doing it
wrong. Now I am using a hack, where I am using prepare() to symlink
the openssl-1.0 files into the build directory and pass them to the
configure script via the --with-ssl-dir option.
If anyone has a cleaner way of doing this, please feel free to share
it. Any help is greatly appreciated.
I also had a look at the openssl-1.1.0 patch the openssh package
from the core repo is using: It does apply cleanly but it is not
enough. I am afraid we will have to wait for upstream to get a
proper openssl-1.1.0 patch.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Works on Raspberry Pi 3 (and probably on Pi 2 as well, as the Pi 3 is
actually using the Pi 2 image for now).
|
|
Seems to depend on "linux-api-headers", rather than "linux-headers". The
former is installed as a dep to glibc (a base group package) and as such
not explicitly listed as a make dep for now.
|
|
Please note, there seems to be a problem with unencrypted transfers >1GB
which is still under investigation upstream, see:
https://github.com/rapier1/openssh-portable/issues/5
|
|
The upstream maintainers fixed some style issues in their HPN code and
also synced their code with the current version of openssh-portable.
|
|
1. PAM was added as a hard dependency
2. openbsd-netcat was removed, as the package built just fine without
ANY netcat version installed. Let me know if I am wrong after all.
3. GDB was removed as the PKGBUILD does not use a check() function
|
|
The function pkgver() depends on the upstream tags. The way this
function was originally implemented seems unable to create strictly
monotonous version numbers because the revision is not properly
increased. Whether a version is higher or lower depends solely on the
commit hash which is somewhat random. I suspect part of the problem
might also be the way the project is tagged upstream, possibly
overwriting tags. Making the function use only annotated tags seems to
fix the issue for now, but I will re-visit this later, especially since
openssh-git is using the same method for deriving a version number.
|
|
|
|
1. Changed maintainer
2. Improved description
3. Added security disclaimer
|
|
|
|
|