* -*- mode: outline; -*- Time-stamp: "FIX.txt: Fri Oct 4 16:12:58 2013"

* DSO 2.7 Build DG17B | 22724

** Improved recorder and display efficiency | 19494 04
   Previously the data recorder listened to all capture data types and
   all enabled channels and discarded the ones it did not need for the
   recording afterwards. This change modifies DDR operation to avoid 
   listening (and processing) all data types not currently enabled to
   record which reduces CPU load (especially important for Raspberry
   Pi). Likewise the display engine no longer calculates and discards 
   data but instead ignores it in the first place if it's not needed.

** Reduced DDR file verbosity when applicable | 19494 03
   When recording raw, enhanced or decimated data there is no wide
   band or glitch data captured to record. This bug caused it to be
   recorded anyway in some cases wasting bandwidth and creating much
   larger DDR files than were necessary. Fixed. 

** Fixed bug preventing logic D1~D7 from recording | 19494 02
   In some situations the DSO Data Recorder failed to record logic
   channels D1~D7. Logic D0 was recorded and the analog channels as
   well but when the bug was activated the additional logic channels
   would sometimes not appear in the recorded output or replay on the
   display in REVIEW or REPLAY modes. Fixed.

** Fixed regression with CURSOR button on some models | 22726 01
   When DSO connected to some BitScope models the CURSOR button would
   become inactive when certain virtual instruments were selected (e.g.
   selecting MIXED mode). Cursors remained active (if they were enabled)
   but they could not be disabled unless a different virtual instrument
   that re-enabled the CURSOR button was selected. Fixed.

** Optimized signal statistics calculations for ARM | 19556 13
   DSO evaluates signal statistics including Vrms, Vpp, Vmin, Vmax
   and Vmean as well time related features such as, signal period,
   rise time, slew rate, duty cycle and so on. Some calculations
   are quite time consuming on slower platforms such as ARM. This
   update optimizes their evaluation to improve performs on these
   systems (Raspberry Pi in particular).

** Fixed regression in spectrum analyzer cursor variables | 22649 02
   The cursor MARK variables (for frequency and magnitude in dB)
   went missing in a recent update. Now fixed. DA15G

** Fixed font regression in Fedora Linux | 22601
   Fedora (and probably other Linux distros) so not install the
   ms corefonts by default but some of these font where used by
   the application. This has been fixed by replacing these with
   their Linux equivalents.

** Added path reporting to DSO system info page | 12160 04
   The SETUP->System page now reports the locations used by DSO
   to find user, configuration, state, library and documentation
   files. These files' location vary by operating system so this
   information is helpful because it provides explicit path names
   for these files on the system on which the software is running.

** Relocated probe file to XDG standard location | 12160 03
   The default location for the local probe file has been updated
   to use the XDG Base Directory Specification. This means the old
   location (for Linux, Mac OS X and Raspbian) is depreciated and
   should not be used. However, if a probe file exists in the old
   location (~/.bitscope/config/bitscope.prb) and not in the new
   one (~/.config/bitscope/bitscope.pdb) it will still be used.

** Built-in probe file defaults now used in Mac OS X | 21734 08
   If the global and local probe file do not exist when DSO is run,
   the local version is created automatically upon SETUP using the
   built-in defaults. Previously the probe file was not always being
   created on OS X requiring it to be manually created if the built-
   in defaults were not appropriate for a given setup.

** Fixed subtle probe file parsing bug | 18838 12 13
   If the probe file contained blank lines (technically illegal) it
   was possible to bomb the parser if they appeared at the end of
   the file. This change makes such lines legal (but redundant) and
   ensures that the library does not bomb when they are encountered.

** Fixed potential race condition in network links | 22415
   While never seen in our testing in the lab this bug had the potential
   to hang the application when talking to network devices. Fixed.

** Fixed minor memory leak upon device reconnection | 22398 03 04
   DSO leaked a small amount of memory each time the application was
   reconnected to a different device. Fixed.

** Improved slow clock trigger point precision | 22037 05
   The trigger point is now calculated with greater precision when
   capturing on some mode BitScopes that use the slow clock capture
   mechanism. This includes BS300, BS310, BS320 and BS50. The newer
   model do not need or use slow clock capture.

** Improve Linux 3.2.x kernel initialization reliability | 21656 14
   BS1xx models require a slightly modified connection sequence in recent
   editions of the Linux kernel ensure they connect first time every time.

** The DSO capture engine has been optimized for performance | 21710
   The capture engine at the core of DSO is now shared by the BitScope Library.
   This means improvements made for the Library are inherited by DSO and vice
   versa. First up, DSO inherits frame rate performance improvements developed
   for the library (BitLib 1.5 DA21D and later).

** DSO logic trigger LED now correctly indicates status | 21656 13
   Previously the LED did not differentiate between trigger waiting and
   trigger fired (instead reporting the same thing until the capture had
   completed). This change makes it clear when waiting for trigger so it
   is easy to see whether one is waiting or capturing.

** Updated for compatibility with newer Linux systems | 21656 15
   Newer Linux distributions use more recent Linux kernels and driver layers
   and these offer some performance advantages and come with new requirements. 
   These have been accommodated in DSO and tested on the latest Ubuntu releases
   (and Debian Wheezy). Display updates in particular now use much less CPU to
   perform at the same refresh rate as before.