[ANN] Vortex Library 1.0.15 'The attack of the mutant protocols!'
Francis Brosnan Blazquez
francis at aspl.es
Fri Jul 4 18:38:03 PDT 2008
## Vortex Library 1.0.15 ##
## "The attack of the mutant protocols! release" ##
Advanced Software Production Line is proud to announce the Vortex
Library 1.0.15 "The attack of the mutant protocols!" release
availability: an open source "BEEP Core" protocol implementation
written in C, mapped onto TCP/IP.
Vortex Library has been developed inside the Af-Arch project as its
data transport layer.
Vortex Library is a stable and robust application protocol
development toolkit. Currently it is running under Microsoft Windows
and GNU/Linux platforms.
Check out the following links to get ready with Vortex Library:
[Vortex Library homepage]
[Sources and Win32 binaries]
[Af-Arch homepage where Vortex Library is being developed]
[BEEP Core community web site]
[Axl Library, XML 1.0 implementation]
The release in short
Several bugs fixed and integrated into the regression test to
reproduce and check associated fix. General documentation update.
General BEEP engine update: support to read frame fragment on
greetings (low latency connections) and inside the BEEP header, to
control how is segmented a message via application handlers.
Mayor update to improve memory handling, reducing its requirements.
General update to fix wrong MIME implementation. Added real support
(a more capable API) to handle MIME (access to body content and
All fixings and improvements were also ported into Vortex Library
- Jens Alfke (provided several suggestions and patches to fix bugs).
- Balaz Tirpak (provided patches and worked on vortex-sasl-rename-01
- Mihail Dmitrievsky (provided complete description to fix a bug at
the channel module).
- Benoit Amiaux (reported wrong deallocation handling at sasl
Changes since 1.0.14
* [fix] Moving two log messages from critical/warning to warning/debug
(closing connection wrongly configured at 1.0 and a log when the
channel is stalled). Reported by Jens Alfke.
* [fix] Applied a fix to check channel reference received at channel
creation, under situation where it was denied by the remote side and
the threaded mode is activated. Reported and fixed by Jens
Alfke. Imported fix into 1.1.
* [fix] Updated regression test to reproduce problem fixed by previous
log and check the fix introduced. Making regression test introduced
to check error codes returned by channel creation failures (profile
not supported: 554 and channel creation denied: 421). Update applied
to 1.0 and 1.1.
* [fix] Updated vortex_channel_new documentation to state that the
channel reference can't be used until the process finish.
* [fix] Fixed const declaration at vortex_sasl_accept_negociation and
vortex_sasl_accept_negociation_full. Reported by Dexter.
* [fix] Fixed wrong internal handling for reply sent on a channel
which wasn't checking if the reply sent was complete causing to
signal more replies sent than actual value. Updated regression test
to reproduce and check this problem that was causing a peer to not
be able to close a connection after sending huge replies. Fixed
introduced in 1.0 and 1.1. Problem reported by Mihail Dmitrievsky.
* [new] Moving from internal API to the public the following function
that allows to wait if all replies were sent for a particular
channel. API added:
* [fix] Moving all files to utf-8.
* [fix] Committing change notification vortex-sasl-rename-01
(Proposed with Patch by Balazs Tirpak).
* [fix] Applying change vortex-sasl-rename-01 provided by accepted
change notification (Balazs Tirpak).
* [fix] Updated vortex_connection_shutdown to better reflect its
functions and how it interacts with the side that created the
connection and its associated reference counting.
* [fix] Added new regression test to check wait reply
function. Nothing to fix for now. Updated vortex_channel_wait_reply
function to explain some details and to state that
vortex_frame_unref must be used to deallocate frame returned. Ported
changes into 1.1 series.
* [new] Added new function to vortex frame module to check current
reference counting status. Ported changes into 1.1 series.
* [fix] Renaming internal type GsaslData into VortexGsaslData to avoid
type confusion. Checked all calls to axl_free and replaced those
that require gsasl_free to be used (to ensure the allocation engine
used from gsasl is the same used to release resources). Reported by
Benoit Amiaux. Change introduced into 1.0 and imported into
1.1. Check regression test.
* [fix] Updating internal vortex sequencer implementation to support
any frame size to split into. It was only supported to splitting up
to fill all the available receiver's buffer. Change applied to
* [fix] Fixed vortex initial greetings processing to support reading
greetings fragmented (rather assuming a single shot). Change applied
to 1.0/1.1. Bug reported by Jens Alfke.
* [fix] Committing missing change that implements updates at the
vortex sequencer module.
* [fix] Updated vortex sequencer module to acquire a reference to the
channel associated to a pending transfer and to release it once
finished (fixed race condition on high load data send followed by a
close). Ported changed into 1.1.
* [new] Updated vortex sequencer to allow handling any size of frame
splitting and to check if the channel allows more content to be sent
rather putting into sleep waiting for the next SEQ frame. Updated
vortex sequencer internals to call to a external function that
defines how many data is to be sent on the next frame
produced. Using new function:
vortex_channel_get_next_frame_size. Ported change into 1.1
* [new] Updated vortex channel and handlers to include support to
configure a handler that allows to control how many data must be
used to produce next frame at the sequencer. This will allow to
implement sender patterns that better fits low latency or high
bandwidth connections. Ported change into 1.1. API added:
- vortex_channel_set_next_frame_size_handler (configure)
- VortexChannelFrameSize (handler)
* [fix] Updated vortex_channel_ref to check if the channel is opened
and if the ref counting is ok. Ported fix into 1.1.
* [fix] Making vortex channel to clear its reference to the sequencer
data before posting fixing a possible race condition. Ported fix
* [new] Updated vortex regression test to include initial
implementation to check vortex bandwidth performance. Updated
regression test options to allow disabling timing checks (useful
when running regression test with valgrind and similar
tools). Ported update into 1.1
* [fix] Update into the listener side connection accept engine to
support frame fragmentation on greetings. Updated regression test to
simulate and check this new feature. Ported into 1.1 series.
* [new] Updated vortex connection module to include new functions to
provide and get a frame segmentation which will operate globally to
all channels if no segmentator is found on the particular channel.
Also added a global segmentator support which is used in the channel
no segmentator for the connection and the channel is found. API
Changes introduced in both series (1.0/1.1).
* [new] Added a new function and internal modifications to implement
channel window change. Patch provided by Jens Alfke. Applied
modified and updated regression test to check transfer rate and the
new API introduced. Ported API updated and regression test to 1.1
series. API added:
* [new] Added support to regression test to allow executing a single
test selected. Changed ported to 1.1 series.
* [fix] Separating part of test_02g into test_02h. Changed ported into
* [fix] Updated vortex_channel_is_ready documentation to signify it
only cares about pending replies for messages sent, not those
* [fix] Changing function prototype to return (int) for
* [new] Added new error symbol (VORTEX_EAGAIN) to handle EAGAIN error.
* [fix] Making vortex_frame_send_raw to be more robust against write
operation errors. This was exposed by new window size change support
causing more data to be written. Making vortex_frame_get_next to
also be more robust against frame fragmentation. Several updates
introduced into regression test to reproduce and check fix
* [fix] Updated regression test to check unref after close operation
on a channel. No bug was found. Ported change into 1.1.
* [fix] Making vortex_channel_unref to check reference and internal
reference counting before proceeding.
* [fix] Making vortex_channel_ref to now check ->is_opened
flag. Updating vortex_xml_rpc_invoke to get a reference to the
channel during the invocation process (released later by
__vortex_xml_rpc_invoke). Ported to 1.1.
* [fix] Updated documentation (fixing typos).
* [fix] making vortex connection establishment to support frame
fragmentation on greetings frame, that is, a partial frame received,
still not complete. Applied fix 1.0 and 1.1. This fixes a bug
exposed while using vortex on high latency connections.
* [fix] Fixed wrong frame more flag checking at
vortex_greetings_process. Fix applied to 1.0 and 1.1.
* [fix] Updated vortex frame module to support content fragmentation
at the BEEP header. This fixes a bug exposed while using vortex on
high latency connections with low bandwidth. Fixed applied to 1.0
and 1.1 Checked against regression test_02f, test_02g and test_02h.
* [fix] Updated vortex_frame_get_more_flag to explain to how use the
function to check for the flag in a proper manner. (1.0/1.1).
* [fix] Making vortex_channel_get_data and vortex_channel_get_data to
check internal reference to the hash used. Updated
vortex_connection_new_empty_from_connection to create an empty hash
rather than moving and nullify the reference. Fixed in 1.0/1.1. Bug
reported by Jens Alfke.
* [fix] Flagging vortex_connection_get_mss to be non-portable due to
windows API limitations.
* [fix] Fixing compilation bug at vortex_connection_get_mss.
* [new] General commit to properly implement MIME support inside
Vortex (committing update into 1.0/1.1). Still some work is required
though regression tests are passed. Now vortex fully support MIME
structured format (still missing multipart) allowing to get all
headers found in an incoming message. Also fixes bug produced by not
introducing a CR+LF prefixing content without MIME headers. Added
initial regression test (test_01d) to check MIME parser. API added:
* [fix] Updating autogen.sh files to make all warnings to be errors.
* [fix] Removing not required checks at
* [fix] Updated MIME internal implementation inside frame, to remove
axlHash usage, replacing it with a custom structure to better fit
* [fix] Updated regression test.
* [fix] Making vortex sequencer SEQ frame received processing to be
pushed at the head of the queue to increase priority (1.0/1.1).
* [new] Updated vortex async queue API to include a function that
allows to push data directly into the first position (next item to
be popped). API added:
- QUEUE_PRIORITY_PUSH (macro)
* [fix] Fixed vortex_connection_close wrong memory access, caused by
calling to the function at the listener side, with a connection in
listener or master mode. Added internal checks to properly unref
when required. (Fixed in 1.0/1.1).
* [fix] Removing wrong condition at vortex_frame_get_next (1.0/1.1).
* [new] Committing files to explain BTF syndrome and how it degrades
BEEP performance. Still working on it.
* [fix] Refactored vortex sequencer internal code to make it more easy
to maintain to allow moving it into new features. (1.0/1.1)
* [fix] Making SEQ frame notification handling to always issue a SEQ
frame if a NUL frame is received (to avoid BTF). (1.0/1.1).
* [fix] Making vortex reader check and issue SEQ frames as soon as
* [fix] Making connection creation to disable Nagle's algorithm by
default to avoid BTF (http://www.aspl.es/vortex/btf.html).
* [fix] Committing updates to the BTF document (1.0/1.1).
* [fix] Making vortex sequencer to reuse an internal buffer to perform
send operations, reducing the allocation/deallocation pattern for
each send (1.0/1.1).
* [fix] Making vortex reader to use an internal buffer for SEQ frames
produced to avoid alloc/dealloc operations for each SEQ frame
* [new] Updated vortex frame API to include new functions to produce
frames on a provided buffer. API added:
* [fix] Updating vortex_channel_set_serialize documentation to better
explain its function and how to use the function at the listener
* [fix] Improved close in transit support by checking more scenarios,
while supporting previous ones. Removed wait reply once an
affirmative reply is found to avoid generating a fake accept when it
is not required. (1.0/1.1). Funny.
* [fix] Updated regression test and vortex engine to support ordered
delivery at the server side (activated by
vortex_channel_set_serialize) which wasn't implemented. Test 02i
reproduce and check the enforced ordered delivery.
* [fix] Moved WaitReplyData structure to vortex channel module to
avoid exposing its details to the API.
* [fix] Making vortex_connection_is_blocked and
vortex_connection_block internal handling to be done with a direct
variable inside the VortexConnection structure to avoid using the
general purpose hash to store its state, improving 4,12% vortex
reader loop. Fix (1.0/1.1).
* [fix] Fixing vortex_frame_get_next to avoid calling to
axl_stream_cmp 5 times to check the BEEP header (missing else if was
causing not required function activations). Fix (1.0/1.1).
* [fix] Making vortex_frame_get_next to avoid allocating twice memory
for each frame received. Now the function reuse the buffer used to
receive the frame ans holder for the VortexFrame structure received.
* [fix] Updated regression test_04a to check different sizes and
number of blocks to transfer. (1.0/1.1).
* [new] Updated vortex profiles API to include two new functions as
part of the automatic MIME header addition configuration at profile
level. (1.0/1.1) API added:
* [fix] Making internal frame MIME state to implement a reference
counting mechanism to share the information between several
frames. Especially used by vortex_frame_copy. Fix applied to
* [new] Implemented library level MIME header automatic addition
configuration through a new item at vortex_conf_set/get. (1.0/1.1)
- VORTEX_AUTOMATIC_MIME_HANDLING (enum)
Updated vortex_conf_set/get to properly support this new value.
* [new] Implemented channel level automatic MIME header addition.
(1.0/1.1) API added:
* [fix] Updated vortex channel module documentation (1.0/1.1)
* [fix] Fixed race deallocation condition at second level
frame received execution. Applied fix at (1.0/1.1).
* [fix] Finished initial implementation with full MIME structure
support. Added regression test to check features introduced
(test_01d). Still more documentation is required (1.0/1.1).
* [fix] Fixed autoconf files to detect mingw platform and add ws2_lib
reference to the linker. Applied patch into 1.0/1.1.
* [fix] Enforcing reference to axl 0.5.3 release.
* [fix] Committing changes to MIME documentation.
* [fix] More documentation about MIME (1.0/1.1).
* [fix] Committing more documentation to explain how applies MIME
support to sending functions (1.0/1.1).
* [fix] Fix bug at vortex_connection_cleanup which wasn't properly
terminating a mutex (connection_new_notify_mutex). It was
initializing it again! (????).
* [fix] Improved MIME regression tests and fixed some bugs under
situations were messages received aren't MIME ready. (1.0/1.1).
* [fix] Fixed wrong documentation associated to
* [fix] Making vortex_frame_get_content_type and
vortex_frame_get_transfer_encoding to return NULL if it is found
a frame without MIME content. (1.0/1.1).
* [fix] Making vortex_frame_get_content to return the payload
(raw message) in the case a non-MIME message was received.
* [fix] Fixed regression test for MIME support on windows (fopen
Advanced Software Production Line is leading the Af-Arch project: a
complete framework to develop distributed application to manage
Af-Arch project relies on Vortex Library to exchange data between
its distributed nodes.
Advanced Software Production Line also provides GNU/Linux support
and consulting services to help organization to introduce GNU/Linux
inside its process.
Contact us, using English or Spanish, to get commercial support
and/or BEEP based development services.
You can reach us:
http://www.aspl.es - info at aspl.es
We hope Vortex Library help you. Enjoy Vortex Library!
Francis Brosnan Blázquez - francis at aspl.es
Advanced Software Production Line - http://www.aspl.es
4th jul 2008, Madrid (Spain)
More information about the BEEPwg