This is an automated email from the ASF dual-hosted git repository.
twolf pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/mina-sshd.git
The following commit(s) were added to refs/heads/master by this push:
new 01d91a6ee Update changelog for 2.9.2
01d91a6ee is described below
commit 01d91a6ee043d309b7c07cb644085a4a696fc32b
Author: Thomas Wolf <tw...@apache.org>
AuthorDate: Tue Nov 15 20:37:28 2022 +0100
Update changelog for 2.9.2
---
CHANGES.md | 49 ++-----------------------------------
CHANGES.md => docs/changes/2.9.2.md | 22 +----------------
2 files changed, 3 insertions(+), 68 deletions(-)
diff --git a/CHANGES.md b/CHANGES.md
index 1e369f47d..58dd8edb9 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -18,61 +18,16 @@
# [Version 2.9.0 to 2.9.1](./docs/changes/2.9.1.md)
+# [Version 2.9.1 to 2.9.2](./docs/changes/2.9.2.md)
+
# Planned for next version
## Bug fixes
-* [SSHD-1173](https://issues.apache.org/jira/browse/SSHD-1173) Not fully using
up a channel window may lead to hangs (see [Channel windows](#channelwindows0)
below)
-* [SSHD-1287](https://issues.apache.org/jira/browse/SSHD-1287) SFTP: reading
with buffers larger than 126kB leads to data corruption
-* [SSHD-1293](https://issues.apache.org/jira/browse/SSHD-1293)
ExplicitPortForwardingTracker does not unbind auto-allocated port
-* [SSHD-1294](https://issues.apache.org/jira/browse/SSHD-1294) Close
MinaServiceFactory instances properly
-* [SSHD-1297](https://issues.apache.org/jira/browse/SSHD-1297) Avoid
OutOfMemoryError when reading a public key from a corrupted Buffer
-* [SSHD-1302](https://issues.apache.org/jira/browse/SSHD-1302) Reading from
Channel.getInvertedOut() after EOF was reached throws IOException instead of
returning -1
-* [SSHD-1303](https://issues.apache.org/jira/browse/SSHD-1303) Reading from
redirected Channel.getInvertedErr() delivers stdout; should be at EOF
-* [SSHD-1307](https://issues.apache.org/jira/browse/SSHD-1307) [NIO2] TCP/IP
port forwarding: shut down output stream only after pending writes have been
written
-
-* [GH-263](https://github.com/apache/mina-sshd/issues/263) Race condition in
BufferedIoOutputStream
-* [GH-266](https://github.com/apache/mina-sshd/issues/266)
ChannelPipedOutputStream.flush() must be a no-op
-
## Major code re-factoring
## Potential compatibility issues
## Minor code helpers
-* New utility method `KeyUtils.loadPublicKey()` to read a public key file.
-
## Behavioral changes and enhancements
-
-* Netty I/O back-end: respect configurations for
`CoreModuleProperties.SOCKET_BACKLOG` and
`CoreModuleProperties.SOCKET_REUSEADDR`.
-* MINA I/O back-end: use `CoreModuleProperties.NIO2_READ_BUFFER_SIZE` for the
initial read buffer size, if set.
- A new `CoreModuleProperties.MIN_READ_BUFFER_SIZE` can be set to control the
minimum read buffer size (64
- bytes by default in Apache MINA).
-* NIO2 I/O back-end: in TCP/IP port forwarding, shut down the output stream of
a socket when a `SSH_MSG_CHANNEL_EOF` message
- is received on the SSH channel only after still pending writes have
completed. See [SSHD-1307](https://issues.apache.org/jira/browse/SSHD-1307).
- The MINA and Netty I/O back-ends already did so.
-
-<!-- --><a id="channelwindows0"></a>
-
-### Channel windows
-
-Previous versions of Apache MINA sshd (from 2.6.0 to 2.9.1) did not always
fully use up a channel window
-and waited for a `SSH_MSG_CHANNEL_WINDOW_ADJUST` message from the peer
instead. They did so if the available
-window size was smaller than the packet size of the channel, and also smaller
than the amount of data still
-to be written. There were settings to change this behavior and always fully
use up a channel window: these
-settings were
-
-* `CoreModuleProperties.ASYNC_SERVER_STDOUT_CHUNK_BELOW_WINDOW_SIZE`
-* `CoreModuleProperties.ASYNC_SERVER_STDERR_CHUNK_BELOW_WINDOW_SIZE`
-* `SftpModuleProperties.CHUNK_IF_WINDOW_LESS_THAN_PACKET`
-
-By default, they were `false`; if set to `true`, the window would be used
fully.
-
-Not using up a channel window may lead to hangs with peers that send the
`SSH_MSG_CHANNEL_WINDOW_ADJUST` message
-only when the window size is very low, or even zero. The SSH RFCs do not
mandate any particular point at which
-an implementation should adjust the window. OpenSSH and Apache MINA sshd
itself do so when half of the window
-is used up, but there are other implementations that do so only when the
available window size becomes zero.
-
-In this version, the above settings have been removed. Apache MINA sshd
behaves always as if they were `true`, i.e.,
-if there is some window space and there is data to write, data will be
written. See Apache MINA sshd issues
-[SSHD-1123](https://issues.apache.org/jira/browse/SSHD-1123) and
[SSHD-1173](https://issues.apache.org/jira/browse/SSHD-1173).
diff --git a/CHANGES.md b/docs/changes/2.9.2.md
similarity index 87%
copy from CHANGES.md
copy to docs/changes/2.9.2.md
index 1e369f47d..5f2d15224 100644
--- a/CHANGES.md
+++ b/docs/changes/2.9.2.md
@@ -1,24 +1,4 @@
-# [Version 2.1.0 to 2.2.0](./docs/changes/2.2.0.md)
-
-# [Version 2.2.0 to 2.3.0](./docs/changes/2.3.0.md)
-
-# [Version 2.3.0 to 2.4.0](./docs/changes/2.4.0.md)
-
-# [Version 2.4.0 to 2.5.0](./docs/changes/2.5.0.md)
-
-# [Version 2.5.0 to 2.5.1](./docs/changes/2.5.1.md)
-
-# [Version 2.5.1 to 2.6.0](./docs/changes/2.6.0.md)
-
-# [Version 2.6.0 to 2.7.0](./docs/changes/2.7.0.md)
-
-# [Version 2.7.0 to 2.8.0](./docs/changes/2.8.0.md)
-
-# [Version 2.8.0 to 2.9.0](./docs/changes/2.9.0.md)
-
-# [Version 2.9.0 to 2.9.1](./docs/changes/2.9.1.md)
-
-# Planned for next version
+# Introduced in 2.9.2
## Bug fixes
