Prep for 0.3.19 release

git-svn-id: svn://svn.code.sf.net/p/nuttx/code/trunk@1319 42af7a65-404d-4744-a932-0658087f49c3
This commit is contained in:
patacongo
2008-11-27 03:18:04 +00:00
parent 2ab40252bc
commit f14b4b3942
3 changed files with 114 additions and 100 deletions
+31 -1
View File
@@ -700,7 +700,7 @@ the NuttShell (NSH). These include:
Several bugs were fixed, the most important of which are:
* Fixd a race condition workaround delay in LPC214X SPI logic. This was also
* Fixed a race condition workaround delay in LPC214X SPI logic. This was also
the cause of some bad MMC/SD performance on that platform.
* Fixed a recently introduced FAT file system problem: It would mount a (invalid)
FAT file system even if the medium is not formatted!
@@ -717,5 +717,35 @@ Please report any errors to me.
This tarball contains a complete CVS snapshot from November 16, 2008.
nuttx-0.3.19
^^^^^^^^^^^^
This is the 31st release of NuttX. This release includes the following new feature:
* Add poll() and select() APIs that may be used to monitor for data
availability on character devices or TCP/IP sockets.
* Implemented support TCP/IP connection backlog. This allows select to wake-up
on new connections to a listener socket.
* Added definition of a framebuffer driver and implement framebuffer drivers
for the simulated platform and the TI DM320 (untested as of the inital check-in).
* Partially developed a graphics framework based on the framebuffer drivers,
however, this will not be ready for use for a few more release. Currently
this includes only a few color conversion routines and some rasteizing functions.
A tiny windowing system is under development but not ready for check-in yet.
* Added support for fixed precision math.
* Added support for outgoing multicast packets.
Several bugs were fixed, the most important of which are:
* Fixed an important bug in the TCP/IP buffering logic. When TCP/IP read-ahead is enabled
and not recv() is in-place when a TCP/IP packet is received, the packet is placed into
a read-ahead buffer. However, the old contents of the read-ahead buffer were not being
cleared and old data would contaminate the newly received buffer.
* Changed the behavior of the serial driver read. It now returns data as it
is available rather than waiting for the full requested read size. This
makes functions like fgetc() work much more smoothly.
These changes were verified only on the Neuros OSD (ARM9) using a Linux development
environment. Please report any errors to me.
This tarball contains a complete CVS snapshot from November 26, 2008.