Revision history for Perl extension Net::Traces::TSH.
0.15 Mon Oct 4 2004
- TCP segments with header lengths smaller than 20 bytes are no
longer included in the tally for 'Total ACKs'. In general,
it may be interesting to count the number of TCP segments with
header lengths smaller than the required minimum (20 bytes).
However, in a recent study I noticed that out of 22,592,477,487
segments with the ACK bit set, only 158 had a header smaller than
20 bytes. It is not clear whether these segments belonged to
protocols masquerading as TCP, or if they where erroneous
TCP segments. Anyhow, by now ignoring such segments, the
following holds always:
Total ACK = Cumulative ACKs + Options ACKs
- Minor code modifications and cleanup
- Distribution cleanup based on suggestions from "Writing Perl
Modules for CPAN" by Sam Tregar (Apress 2002)
- POD updates and editorial changes
0.14 Tue Aug 31 2004
- Data structure %Trace and %Interface refactoring
- converters/tsh2ns2.pl ver. 0.01 released
- Updated applications/tsh2tcpdump.pl
- POD updates and editorial changes
0.13 Mon Aug 30 2004
- Added configure() to ease configuration setting
- process_trace() can now convert a TSH trace to tcpdump and ns2
formats (see module documentation) in addition to collecting
per-interface and aggregate protocol usage statistics. The
parameters passed to process_trace() has been reduced to one: the
TSH trace filename to be processed.
- Updated applications/tsh2tcpdump.pl
- Additional tests released
0.12 Sat Aug 21 2004
- Code updates and minor fixes. Use autouse for Carp.
- POD updates and editorial changes
- Added get_interfaces_list()
- Updated documentation/sample-summary.pod
- Additional tests released
0.11 Sun Jun 13 2004
- Application tsh2tcpdump.pl 0.01 released
0.10 Thu Jun 10 2004
- Code reorganization and simplification
- Added get_interfaces_href() and write_interface_summaries(). One
can now get statistics on per-interface basis in a way similar to
get_trace_summary_href() and write_trace_summary(), respectively
- Added write_summary(), an internal, non-exportable utility function.
- Additional tests released
0.09 Mon Jun 7 2004
- Module can now provide summary statistics on a per-interface
basis in addition to the aggregate statistics. (POD to be added.)
- POD editorial changes
- Renamed %Trace_Summary to %Trace
- Updated __DATA__ section with the latest IANA Protocol Numbers
published (May 12, 2004); allowed for more multi-word protocol
names
0.08 Tue May 4 2004
- Code simplifications in date_of() and process_trace()
- POD updates and editorial changes
- Additional tests released
0.07 Sun Apr 25 2004
- Same code as in 0.06, with minor editorial POD changes
- Distribution fix: CPAN testers reported that ver. 0.07 fails to
pass its own tests on MSWin platforms (see comments below,
ver. 0.06). Since this may lead users to believe that the code
does not work correctly, I decided to skip context diffs between
the distribution-bundled sample output files (t/sample.csv and
t/sample.tcpdump) and the locally generated ones on MSWin
platforms for now, and continue searching for an alternative
0.06 Sat Apr 17 2004
- Distribution fix: The tarball comes with sample output files
generated on the development platform (perl v5.8.0 built for
i386-linux-thread-multi). These files are used by the module
tests to validate that the code works correctly regardless of
platform. For example, t/process_trace.t processes t/sample.tsh
and generates t/sample.tcpdump. When a user installs
Net::Traces::TSH on another machine, the module tests verify that
given the same input (t/sample.tsh) the code produces the same
output (t/sample.tcpdump).
Testing this on Linux and UNIX systems is very easy: simply use
_diff_. On systems that lack diff, the tests fail (ver. 0.04 and
0.05). I'd prefer to use a standard "Perl way" to do this, thus
avoiding to rely on external programs. Unfortunately, at this
time Algorithm::Diff is not a core module so I cannot rely on
that either. Currently I employ 30 tests to check that the code
does produce the same results, so users of all platforms have
assurance that the code works correctly, but a context diff would
make most of them redundant. If the platform has diff, it is
used as well (31st test).
- Added t/sample.csv
- Added documentation/sample-summary.pod, an annotated version of
t/sample.csv
- Additional tests released
0.05 Wed Mar 31 2004
- Distribution fixes: META.yml, t/sample.tcpdump
0.04 Tue Mar 30 2004
- Minor updates in the process_trace()/tcpdump text generation
- Code and POD updates
- Added t/sample.tcpdump
- Added META.yml
- Additional tests released
0.03 Sat Mar 20 2004
- Code and POD updates
- pt.pl 0.03 released
- Added t/sample.tsh trace
- Additional tests released
0.02 Wed Mar 10 2004
- Application pt.pl 0.01 released
- Additional tests released
- POD editorial changes
0.01 Mon Mar 1 2004
- Original public release