Updated ChangeLogs one more time for 1.5.2

[geda-gaf/peter-b.git] / gnetlist / ChangeLog

# Do not edit this file - generated from version control history


commit 546d4dbe0124d3f9421116b9e5a8889cccf97b84
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Mar 28 15:22:19 2009 -0400

    Updated all ChangeLogs
    
    Updated the ChangeLogs by running:
        ./update-changelogs.sh 1.0-20070526..master
    in the toplevel directory

commit 21e8c0b232e08927fd091a1d46413990b4c61358
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Mar 28 15:18:34 2009 -0400

    Updated all dates in the relavent man pages (for 1.5.2.20090328)

commit 859abe69912ba99ca27bb58c378284e9eec56259
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Mar 28 15:13:59 2009 -0400

    Updated all versions to 1.5.2 20090328 36:0:0
    
    Updated by running:
    ./update-versions.sh 1.5.2 20090328 36:0:0

commit c3081930ed224394f645de4bfb51e86fa3e15e65
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Feb 20 23:26:12 2009 +0000

    Remove obsolete references to libgd and libgdgeda

commit 3c0d85eae92d5764aa7dee2899c6a066c3470ee3
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Wed Jan 21 14:11:39 2009 +0000

    Write logs to $HOME/.gEDA/logs/.
    
    Rewrite s_log_init() to put log files in a centralised location. A
    user application (e.g. gnetlist) provides a prefix, and log file is
    opened at:
    
       $HOME/.gEDA/logs/<prefix>-<YYYYmmdd>-<serial>.log       [1]
    
    The serial is found by the following procedure:
    
    1. Read through the directory to find the highest existing <serial>.
    
    2. Starting with that number, repeatedly attempt to open()
       incrementing files with O_CREAT|O_EXCL flags to guarantee that a
       new file is created.
    
    3. If after LOG_OPEN_ATTEMPTS = 5 attempts a unique filename hasn't
       been found, fail.
    
    The logic *intentionally* does not fill in gaps in the sequence -- if
    I delete logs, I still want a new log to appear at the end of the
    list.
    
    [1] The actual location is a subdirectory "logs" of the user config
        directory, which may not be ~/.gEDA on all platforms.

commit fda82eb96e2c785efe91fd59d01f6e7e26b897ed
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sun Jan 18 01:48:58 2009 +0000

    Protect some strings passed to pkgconfig.
    
    Ensure that pkgconfig version requirement strings are protected from
    M4 expansion.

commit ee250c9bd2e88bf352a03d1719603b7d54fe430c
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Tue Jan 13 23:58:30 2009 +0000

    Avoid underlinking of apps due to missing explicit depends
    
    These went unnoticed because the apps were resolving the necessary
    symbols when pulling in libgeda and its dependencies.
    
    Fix applications using GLib and guile within the suite to explicitly
    pull in the appropriate CFLAGS and LIBS. This commit removes some
    LDFLAGS and LIBS from grenum, since they weren't being used.

commit 75d03b030f4e3e491710a409bbcf18cc4a16f300
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Jan 10 23:26:28 2009 +0000

    Remove executable permissions from some data-files in the repository

commit 1645fda42758a837c7ccf1efd32fd3b484e22421
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Jan 10 22:17:44 2009 +0000

    Remove unused GTK dependencies from text-based tools

commit a344ff9c33b9bbb7b1a622b70dcedf393e1de09a
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Jan 10 20:41:43 2009 +0000

    Remove some unused Gdk* typed variables from globals.[ch] across the suite

commit 42a5bebc27f842b1955d9626f2521eea124259a0
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Jan 9 01:04:14 2009 +0000

    gnetlist: Teach netlist traversal code to understand buses (still disabled)
    
    Up until recently, the connection scanning code has rejected some, but not
    all buses. This means that it is unlikely graphical buses formed a useful
    part of anyone's design, since the netlist code in s_traverse_net()
    would not follow connections to another bus segment. The only possible
    bus-pin connectivity has ever been "pin-bus-pin", or "pin-bus-net-...".
    
    This behaviour appears to have been accidental, and would not have been
    easy for a user to take advantage of. The traversal code would not move
    on to a subsequent bus, but would, for example, allow a pin->bus->!bus
    connection, which could then connect onwards. The bus mixed up in this
    would for all intents and purposes behave as a net.
    
    The pin->bus validity bug was actually fixed in libgeda when adding
    support for bus pins. This commit formalises it further, since if
    support for netlisting bus pins were enabled it would otherwise still
    be possible to get a buspin-bus-net-... connection if you tried hard.
    
    Introduce type-checking between pins, nets and buses such that they can
    not netlist where the object types don't match. For buses, the code to
    name a connection based on "netname", "net" or "label" attributes is
    disabled - pending decision as to what constructs we wish to use for
    naming buses. Add a gnetlistrc setting (unnamed-busname ...) which sets
    the default naming (if the connection scanning code were run on bus pins).
    
    This commit lays some of the ground work for shared connection scanning
    code for determining the connectivity of buses and bus pins.

commit a838dece88323eb782b0ac301c5825ed693fbf0f
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Jan 9 00:57:40 2009 +0000

    gnetlist: Disable traversal of bus pins when checking connectivity.
    
    We don't want graphical buses to be netlisted at present. Currently
    their output (although not correct connectivity in all but simple
    cases) can get muddled up with the normal netlist fed to backends.

commit d18bfe1738347a1546f5bf1dd5011fb796176de1
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 8 23:05:32 2009 +0000

    gnetlist: Refactor s_traverse_component() and s_traverse_net() slightly.
    
    Removes some duplicated code by not special-casing the first level of
    recursion when tracing out connections.
    
    The only outward change from this commit should be that nets / pins found
    immediately connected to a pin being inspected get their "n" or "P" print
    in the verbose debugging output. (One of the codepaths being consolidated
    didn't have them before). Unconneced pins no longer get a "p" print, since
    the traversal code isn't executed for them.

commit 2d888258d954c993aa03e4a1beb7b03e62df4d95
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 8 00:45:28 2009 +0000

    gnetlist: Avoid depending on GHashTable ordering for output consistency
    
    If the GLib hash table changes (which it has in late versions of GLib),
    the netlist output ordering changes due to our mis-use of GHashTable to
    list all unique component names without enforcing any eplicit order on
    the enumeration of the hash table contents.
    
    Since ordering may be important to backends, want to have known output
    ordering whilst removing duplicate entrys, but without an explicit sort
    of the refdes list.
    
    This commit introduces an incremental build of the list, first checking
    the hash table for a hit, and only adding the new refdes if it wasn't
    already in the has table. This is gives the output a predictable order
    (based on the input), rather than inserting all refdes into the hash
    table, and building the list in whatever order the table implentation
    traverses them in.
    
    This change does unfortunately means a large number or ordering related
    changes are also made in the gnetlist test suite's golden files.
    
    There are also some non-ordering changes in the checked in golden files.
    These is encountered in the pretty well obsolete PCBboard backend - which
    errors out when it hits certain components it can't find footprints for.
    Changing the ordering changes the point at which it dies, and thus the
    netlist output. Lets just accept these changes as is.

commit d57fcbf52893127f5a37656f65c6f076c6d51470
Author: Maciej Pijanka <agaran@pld-linux.org>
Date:   Thu Jan 8 00:33:25 2009 +0000

    gnetlist: Fix test/hierarchy*/Makefile.am to detect failures
    
    Unfortunately the exit code weren't getting propagated back from diff,
    so the test-suite would continue on inspite of errors.

commit d7b7b23721b4054172587c9961d4b87f47bee198
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Dec 31 17:26:03 2008 +0000

    libgeda: Remove non-private access of PAGE->object_list
    
    This allows us to keep better track of where the PAGE must be updated
    should we have any spatially keyed data-structures in the future.
    
    Rename PAGE->object_list PAGE->_object_list to try and discorage any
    use of it outside libgeda (ideally, outside s_page.c) in the future.

commit ca6556ace0a06ec1dd697852f7a1f0ffb86bfb97
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Dec 31 17:25:57 2008 +0000

    Add "const" to some places where we deal with GLists of OBJECTs.
    
    These GLists themselves are not (and shouldn't be) modified, although
    the OBJECTs they point at may well be.

commit 714580df3fee12732d5e1180722f008a47aaf973
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Dec 26 21:53:42 2008 +0000

    gschem: Don't define key-bindings for keys which have no-action
    
    If we do this, then when we set "no-action" to a menu item's key binding,
    the code will try to display one of the keys associated with "no-action"
    as a binding for that menu item.

commit 3c49636e1a9e9e703fab4c6aafc69affb4fd2bab
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Dec 21 16:23:26 2008 -0500

    Updated ChangeLogs one more time.

commit 2f3e07a1fa7fd92ab163d2cd966efd98ab306512
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Dec 21 14:40:21 2008 -0500

    Updated ChangeLogs again in prep for the 1.5.1 release

commit 76d02f7f935d930d8093bd49b7eb0836d3b1aa6e
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Dec 21 14:06:17 2008 -0500

    Updated all ChangeLogs in prep for the 1.5.1 release.
    
    The update was done by running:
    ./update-changelogs.sh 1.0-20070526..master

commit 680e08085d44c9876bdadef0ca7ef25b474c2006
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Dec 21 13:39:29 2008 -0500

    Updated all versions to 1.5.1 20081221 35:0:0
    
    Updated by running:
    ./update-versions.sh 1.5.1 20081221 35:0:0

commit 4f9f4336a123f91318ec2d1ffd8e9739eb756348
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Dec 21 13:10:23 2008 -0500

    Updated dates and versions in all man pages

commit d8cc920d932f1557f3d93da6abd1d99e690f39ef
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 21:47:12 2008 +0000

    gnetlist: Optimise connection traversal algorithm
    
    Increase performance for traversing schematic connections by
    optimising tracking of how many times objects have been visited.

commit 8b171391d26ee79ecccff732ff9ccb257460b815
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 20 16:06:27 2008 -0500

    Fixed up drc2's gnetlistrc to work with both make check and make distcheck

commit edf990235cac2fe6ac2ca6e120f198fd14d05ed7
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 20 14:56:02 2008 -0500

    Added missing liquidpcb to the subdirs in order to make distcheck work better

commit 45dae54fa28ec09313806a80781f9dd6b8d35485
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 20 14:42:40 2008 -0500

    Ignored a few machine generated scripts

commit 24d24fcc3ab6ecbec321d38fc011513df521c122
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 20 14:40:42 2008 -0500

    Fixed a typo in tests.list which caused the JD_nomunge test to not run
    
    Fixing this typo also required the regen of all Makefile.am, creation
    of JD_nomunge.retcode, and changed a few golden files.

commit 59a09888c3a0f70af8312da604a98cb2d0c411d5
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Dec 20 19:36:09 2008 +0000

    gnetlist: Fix regen_sub_makefile.sh
    
    Needs /bin/bash, and an -e argument to the second echo.

commit 581f8cec3d5e717fa78417772759b4525a52f458
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 18:56:35 2008 +0000

    gnetlist: Restore bom2 ordering
    
    The changes in 51c2a5c9f7f9ad1c99b9da2774a051ae97f3e62e caused the
    ordering of components in the BOM output to be reversed. This changes
    the ordering back to the way it was before.

commit 51c2a5c9f7f9ad1c99b9da2774a051ae97f3e62e
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 18:23:22 2008 +0000

    gnetlist: Improve bom2 backend performance.
    
    Change part of bom2 backend to use tail recursion, avoiding stack
    overflows for very large schematics. Tested with up to 1024 devices,
    but gnetlist core performance made testing with larger schematics
    problematic.

commit fff0850cbff67ded4dc53138679e2f5f445f86cd
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Dec 20 17:08:50 2008 +0000

    Revert "gnetlist: Update golden files for backends which hit errors."
    
    This reverts commit 9928a19d3374c5d7ff672f8435a5f29476b2bba2.
    
    gnetlist should not exit with a status of 0 when encountering
    an error. This behaviour change has been fixed in commit
    16102ef095c959b5c1febb9b9259dda23c739258, which reverted part
    of commit 2a4fdb13021d0153e788fe3b2fc005f273dcdf4b, where the
    changes were first introduced.

commit 9904adcf923855f10a64c5fb15cde22d0766155d
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 16:44:01 2008 +0000

    Fix potential buffer overflows in gnetlist.
    
    gnetlist's main_prog() had several uses of sprintf() into a fixed
    length buffer with unchecked user input. Replace these with
    g_strdup_printf() and g_build_filename().

commit 16102ef095c959b5c1febb9b9259dda23c739258
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 16:07:03 2008 +0000

    Quick fix for gnetlist exit status.
    
    In 2a4fdb13021d0153e788fe3b2fc005f273dcdf4b, gnetlist was modified to
    use g_scm_c_eval_string_protected() to call into
    backends. Unfortunately, this had the side effect of stopping Guile
    errors thrown by backends from causing gnetlist to exit with non-zero
    exit status.
    
    This is a quick fix which partially reverts the offending commit.

commit 6329c5a7bdae9573f01f92403e4645e79e66aab5
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Dec 20 14:41:05 2008 +0000

    Update copyright header dates to 2008.
    
    (And I'll have to do this all again in a few days time).

commit 9928a19d3374c5d7ff672f8435a5f29476b2bba2
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Dec 18 20:54:25 2008 +0000

    gnetlist: Update golden files for backends which hit errors.
    
    Gnetlist now returns with an exit code of 0, even when it encounters
    problems from inside the backend's scheme code, and that code exits
    with an (error ...) call. Update the golden files to reflect this.

commit d13aadcbac4af912e0555b3696b59fc904c6cd9f
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Dec 18 20:54:02 2008 +0000

    gnetlist: Checkin missing change to systemc backend
    
    This part of the fixes for the systemc backend got missed from
    commit #bfd49ad477a49235c3e5ee1eda54e5009f6e347a

commit 4ff7f410e4e601b85e5b6d0b8d4b6b930e49072b
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Dec 18 20:51:48 2008 +0000

    gnetlist: Update common test-suite outputs for "systemc" backend
    
    Updated after fixing the backend to run again.
    
    NB: A cursory inspection of the "golden" files, suggests that the outputs
        being checked in are not actually correct. However, the test-suite
        will serve to tell us if any changes are made in gnetlist which affect
        this backend.

commit bfd49ad477a49235c3e5ee1eda54e5009f6e347a
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Dec 18 20:44:59 2008 +0000

    gnetlist: Fixup systemc backend
    
    Don't escape < and > characters in strings, as guile doesn't like this.
    
    Apply similar fixes as those made to the VHDL and verilog backends after
    some changes in gnetlist to accomodate slotting in spice-sdb. Since the
    netlist backend is loaded before gnetlist has traversed the schematic, it
    must not execute any code which queries gnetlist on load. All such work
    is to be done only when gnetlist invokes the "systemc" method.
    
    Move (define c_p #f) to outside (systemc:components ...)'s (lambda ...)
    function, since guile doesn't appreciate us mixing declarations and
    expressions. This variable is tested and (set! c_p #t) later on, which
    doesn't feel very scheme like. The fix works for now though.
    
    Check to see if the regexp matched netname, checking for "name<type>" has
    hit a match before trying to get a piece of that match. If the regexp
    returns #f, instead just treat the identifier as a "name" and use that.
    This allows net names which don't conform to the regexp to at least give
    some vaguely sensible output, without crashing the backend.
    
    Comment some debugging prints to stdout, giving a cleaner looking run.

commit 55850bb331afc0596a18dac1807a487cb475b30d
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Dec 18 19:05:47 2008 +0000

    gnetlist: Add new "liquidpcb" backend to the common test-suite

commit 617797b5a47d370c4fee19110c826502b7e3082b
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Tue Dec 9 16:19:09 2008 +0000

    Remove linkage between OBJECTs and use GList for OBJECTs on the page
    
    This affords quite a lot of code clean-up, since we no longer have to
    maintain two versions of each operation which works on lists.

commit d3a0d4da489f9cd213472e58d18f21774760984b
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Mon Dec 8 17:57:17 2008 +0000

    Tolerated changes in golden files for the spice netlist backend.
    
    The way slotted parts are handled has changed since 1.2.0, as
    has the output of the spice backend for slotted parts. Since
    the spice backend doesn't properly support slotted components,
    we allow these output changes.

commit 227e5102e722675c93ecf897b1b1c422dd7f032c
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Mon Dec 8 17:48:06 2008 +0000

    gnetlist: Revert removal of spice backend testing
    
    This backend was removed from the common test-suite prior to the last
    release due to failures. We still want to test the backend, although
    we may wish to accept some changes to the golden files before the next
    release.

commit 40e4a0836529f91ee141463832beb4fea0d845ec
Author: Dan McMahill <dan@mcmahill.net>
Date:   Mon Dec 1 20:34:02 2008 -0500

    Use the uninstalled scheme files instead of installed scheme files.
    
    When running the tests, point at the not-yet-installed scheme files.
    This lets 'make check' test out what you just built instead of something
    that is installed that may be sorely out of date.

commit 817e5dcfe478ae02f60d74ca1574518a1976bd17
Author: Dan McMahill <dan@mcmahill.net>
Date:   Mon Dec 1 02:55:24 2008 -0500

    Run all tests and report failures at the end.  Also add missing .PHONY.
    
    - Add missing .PHONY to the tests target.
    - Rather than exiting on the first test that fails, run them all and
      keep track of how many failed and which ones.  This lets you run
      through all the tests and then come back to the failed ones.

commit 635546dcc9615c490e85ff6e7ef0c7d990af6842
Author: Dan McMahill <dan@mcmahill.net>
Date:   Sun Nov 30 00:04:02 2008 -0500

    Specify the path to the gnetlist binary being tested.
    
    Point at the locally built but perhaps not yet installed gnetlist
    when running 'make check'.  This helps make sure we are testing
    the right gnetlist.

commit 51896e033129dada3f96a3448e189e19554c239a
Author: Dan McMahill <dan@mcmahill.net>
Date:   Sat Nov 29 22:11:32 2008 -0500

    Remove files which are now generated at configure time from templates.
    
    These files are now generated from the .in versions at configure time
    to make sure we get a suitable awk implementation called out.

commit 1625b379e7066ffe2a8283a4977c794ff55b7db4
Author: Dan McMahill <dan@mcmahill.net>
Date:   Sat Nov 29 21:11:07 2008 -0500

    General portability cleanup of scripts.
    
    Expand the autoconf tests for awk to make sure that we have picked up and
    awk that is good enough for the scripts that use them.  Now use this result
    in various scripts instead of hard coding 'gawk' (not available on all systems)
    or 'awk' (picks up the original awk which is not sufficiently advanced on solaris).
    
    Also avoid calling out /bin/bash on scripts for which /bin/sh (even on solaris) should
    be sufficient.  Not all systems have /bin/bash.
    
    Finally, improve temp file usage in shell scripts for improved security.

commit 1d6a2045f618cb2441d386811a49fc3c1686ff13
Author: Dan McMahill <dan@mcmahill.net>
Date:   Sat Nov 29 20:54:19 2008 -0500

    General portability cleanup of scripts.
    
    Expand the autoconf tests for awk to make sure that we have picked up and
    awk that is good enough for the scripts that use them.  Now use this result
    in various scripts instead of hard coding 'gawk' (not available on all systems)
    or 'awk' (picks up the original awk which is not sufficiently advanced on solaris).
    
    Also avoid calling out /bin/bash on scripts for which /bin/sh (even on solaris) should
    be sufficient.  Not all systems have /bin/bash.
    
    Finally, improve temp file usage in shell scripts for improved security.

commit 5bf126369c5f5a60f5ae039170c33ff4c767a4ee
Author: Dan McMahill <dan@mcmahill.net>
Date:   Wed Nov 26 22:48:08 2008 -0500

    Avoid insecure temp file usage.
    
    Fixes the security vulnerability noted in http://secunia.com/advisories/32806/
    The issue is insecure temp file usage.  The fix is to create a private directory
    and keep temp files in there.

commit c7dfc632ea75a802822fc139a93a96a3dfdcd1b2
Author: Dan McMahill <dan@mcmahill.net>
Date:   Tue Nov 18 22:47:00 2008 -0500

    When testing gnetlist, point to the one we just built.
    
    Make sure that 'make check' works correctly prior to installation.  The
    old test seemed to be picking up a previously installed gnetlist instead
    of the one we just built.  Makes for some confusing debugging if there is
    another version installed and in your path somewhere.

commit f470af7a28afe423ded6171ea3c8a2b31ff56f43
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Nov 1 20:47:41 2008 +0000

    Revert applying -Werror to compiler flags when linking with GCC (CODE!!)
    
    Depending on the gcc version, the original developers might not see the
    warning until the software gets into the end user's hands (and tries
    building it with an older/newer version of gcc).  Most warnings are
    not catastropic, but by using -Werror they are _all_ catastropic.
    
    If a developer wants to get rid of all warnings, they should either:
    
     1) run make -s and see/fix all the warnings, and/or
     2) modify the Makefile's themselves and add the -Werror flag temporarily.
    
    (Comments explaining rationale snipped out of an email from Ales).

commit 3a748314907a0167fc06fe08be498f40c6c7f318
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Nov 1 17:46:01 2008 +0000

    More placeholders for fixing some warn-unused bugs
    
    Some were missed in the last (accidental) commit
    relating to these warnings.

commit 2b1c2026611ef56fba3c07e88d1292b19047e818
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Nov 1 17:45:57 2008 +0000

    Use -Werror when building with gcc compilers. (More Makefile.am)
    
    Hopefully this will help to ensure compiler warnings get noticed
    and fixed by developers.
    
    Some Makefile.am with compiled sources were missed in the last
    (accidental) commit relating to these CFLAGS.

commit 09eacbfb36e2b00758bce8d8add2f8f56cdcfdbb
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Nov 1 16:09:26 2008 +0000

    Use -Werror when building with gcc compilers.
    
    Hopefully this will help to ensure compiler
    warnings get noticed and fixed by developers.

commit 816c8735215b76faeaa15bfc5de01d345aedc87d
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Oct 26 20:07:13 2008 +0000

    Update golden files for gsch2pcb backend
    
    These were out of date since changes were made to make the gsch2pcb
    backend to emit boards which fit better with PCB's current default
    options and layer stackup.

commit aa6b1cf719bc8c31c1763a6160e133e6e3bc46ce
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Oct 26 20:07:12 2008 +0000

    Update golden files for PCBboard backend
    
    The "singlenet" output now emits the PKG_DIP14 package correctly,
    whereas the previous golden file just has the M4 macro listed
    unexpanded.
    
    Whether some bug was fixed, or if this was just a mistake when the
    previous golden files were generated is unknown, but the new output
    is better.

commit 2b5ac5d34a71bfb5f91012ac14bbf2e442858303
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Oct 26 20:07:12 2008 +0000

    gnetlist: Update golden files for desired BOM2 changes
    
    BOM2 now has a qty column, so update the test-suite to match.

commit bcaf9f8dafcd25335b61b2ea80d071bde720bbd1
Author: Cesar Strauss <cestrauss@gmail.com>
Date:   Sun Oct 26 10:41:31 2008 +0000

    Use pkg-config support for Guile if present.
    
    Using pkg-config allows easier cross-compilation, since the build
    process can use a pkg-config binary native to the build machine with .pc
    files for the host machine.

commit 2a4fdb13021d0153e788fe3b2fc005f273dcdf4b
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Tue Oct 21 16:52:09 2008 +0100

    Validate calls to scm_c_eval_string(). [2105219]
    
    Because the reporter's version of Guile is broken, a lovely garbage
    collector segfault occurs if a null string is passed to
    scm_c_eval_string().
    
    For now, wrap all calls to scm_c_eval_string() with mutable arguments
    in g_scm_c_eval_string_protected(), and since the latter function
    seems to behave when passed a NULL string this bug should just go
    away.

commit 79ab7bd82c3f2bb5266a6065512730157798ec2d
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Mon Oct 13 21:44:38 2008 -0400

    Added first cut of the LiquidPCB (http://www.liquidpcb.org) backend
    
    Also added a couple of trivial tests for this new backend
    as well as updated the man page for gnetlist.

commit f0a96de8b545dbd900e87db6abf54a0b6df830ab
Author: Werner Hoch <werner.ho@gmx.de>
Date:   Sat Oct 11 09:02:03 2008 +0200

    .gitignore: log files and source tagging files added
    
    Added the files of the "global" source tagging program
    and the logfiles of the gEDA programs to .gitignore.

commit 2a9f5584219a894557aa7fd97f94a9a941c5bc1b
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Oct 1 18:17:19 2008 +0100

    gnetlist / gsch2pcb: Update default board to match current PCB version.
    
    Ensures greater compatibility with layouts created using PCB from the
    outset, rather than via gsch2pcb.

commit a17b67c07298e707fbfa2d10f8ec18c16b0b7482
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Oct 1 17:57:42 2008 +0100

    gnetlist / gsch2pcb: Fix default PCB board to have lines clear polys.
    
    This bad default keeps biting users of the gsch2pcb work-flow.

commit 05d7c0a430e2334fc24f561268f5dbcee62d14ad
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Sep 24 02:37:02 2008 +0100

    gnetlist: Tidy vams_get_package_attributes() and vams_get_attribs_list()
    
    Refactor the code to be cleaner, fixup whitespace, remove old debug cruft.

commit fbdc2486cc9bf99bc79408ef5b08189a94c89bc1
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Sep 24 02:36:22 2008 +0100

    Remove ineffective search inside symbol in vams_get_package_attributes()
    
    The call to vams_get_attribs_list (...->complex->prim_objs, ...) will
    always fail to return names. This function operates on the passed
    OBJECT, looking at OBJECT->attribs. In this case, the passed OBJECT
    will always be a complex head node with no attributes.
    
    We _might_ want to list the names of any floating attributes inside the
    object (as seems to have been intended in the original code), however
    all "interesting" generic attributes will probably have been promoted to
    the outside anyway. Promotable attributes inside symbols are only hidden
    by default (not deleted), so listing the internal attributes would mean
    we end up with two copies of each promotable attribute in our SCM list.
    
    There are also quite a number of uninteresting attributes inside most
    symbols which we may not want listing as generics. For now delete the
    broken code attempting to list internal attribute names.
    
    This commit doesn't change any netlist output.

commit f4dab83b1ca0e197f02f520a6ae525aeac5a252d
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Sep 24 01:07:43 2008 +0100

    gnetlist: Use the complex_parent property of an OBJECT, not the HEAD node.
    
    This saves having to traverse back to the HEAD node for each lookup.

commit b4ed5e92d203c7449b7d33e924a41b093e1de930
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Tue Aug 26 12:55:07 2008 +0100

    Remove duplicate definitions of do_logging.
    
    Also sets the default in libgeda to TRUE.

commit c95bd0a1f21be899510e07af28d0190fc09dc469
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Aug 24 22:27:29 2008 +0100

    Remove unused #define DEFAULT_SERIES_NAME ... defines

commit a974a4c6d2c233c9209c5b4597f7d4f86154f484
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Aug 24 22:27:28 2008 +0100

    Remove default-series-name from all rc files
    
    Fixes brokenness leftover from previous removal of "series_name".

commit 73a73cd0cb54d0f9430075aa00359b7840bd8d2c
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Aug 17 02:19:49 2008 +0100

    Add qty column to bom2 gnetlist backend
    
    It is convenient for board-assembly houses who want to see a qty.
    against each line-item they have to procure.

commit e36585def8789540f25464edecbf119706882958
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Aug 17 01:59:29 2008 +0100

    gnetlist: Don't drop all hierarchy prefix for children without refdes=...
    
    If a hierarchy block is missing its refdes=... attribute, insert it
    without any change of hierarchy prefix. The block's contents keep the
    same hierarchy level as the instantiating schematic. Previously the block
    was inserted at the root level, with no hierarchical name mangling.
    
    This may be useful in some cases if you want to suspend hiearchy mangling
    for certain portions of a hierarchical design. (NB: You would then be
    responsible for avoiding refdes name clashes). A similar effect can be be
    obtained on a design wide basis by setting the options:
    (hierarchy-*-mangle "disabled") in a gnetlistrc file.

commit 1a1460c65a9d4e0aee35e39fff0f083c95bd380d
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Aug 16 19:18:24 2008 +0100

    Remove the ATTRIB struct as it only contained one variable.
    
    The OBJECT.attribs GList now points directly at the text objects which
    hold and display the attribute name and value.

commit 11a2692a1ab8bcee14178286fd719d1ec39e6b9a
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Aug 16 15:24:16 2008 +0100

    Draw func cleanup.
    
    Rather than having libgeda require other applications to provide the
    *_draw_func symbols, libgeda should provide them itself. Other
    applications should then override them as required near their call to
    libgeda_init().

commit 723e8d1c735b27f2bfc39dec296464bb8d5b2020
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Thu Aug 14 23:27:12 2008 +0100

    Don't register recalc funcs as draw funcs.
    
    Some applications which don't draw schematics set the draw funcs to be
    the recalc funcs. This is silly.

commit 57a4fd272e10a68e05051a9ab9309e252e2103f7
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:03:12 2008 +0200

    Purge AC_CHECK_FUNCS() in configure.ac in each package of unused functions.

commit 872287f23b4880b01babed777efe7976783feab2
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:03:05 2008 +0200

    Cleanup of libgeda o_attrib_get_name_value().
    
    Rewrote o_attrib_get_name_value(): it now has support for UTF-8
    attribute strings ; plus it can return only the name or the value or
    both name and value.
    
    Finally adapted code wherever it can benefits this new feature.

commit a76cf5cdda91c7070f4765107d3f8be89bbd05b6
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:03:01 2008 +0200

    Remove unnecessary strippings of trailing directory separator.
    
    These strippings is no more necessary as GLib functions for building
    filenames (in particular g_build_filename()) takes care of any
    trailing directory separator for us.

commit cd70e88b8c7faa4beea8b13b846e92807a252e63
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:03:00 2008 +0200

    Replace getcwd() with g_get_current_dir().

commit 90b0ece453efc80490c6c277386f4f5f7bfdc4af
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:02:59 2008 +0200

    Simplify build of filenames with g_build_filename() (!).
    
    Modified code where possible to use preferably g_build_filename()
    instead of g_build_path(), or g_strconcat() and friends.

commit 19d6d561eb5aea8e580d6ecf72d20bdfc3180ace
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:02:58 2008 +0200

    Remove unnecessary check following memory allocation with GLib g_malloc().
    
    There is no need to check g_malloc() returned value as when the
    function fails to allocate memory, it terminates the application.

commit a3513b7e6dc7bf6c36d95d4da86082beeb79831d
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:02:57 2008 +0200

    Replace standard string manipulations with GLib string utility functions.

commit da1d2e7b304f9d832bdee9c8dfb9fe4604e19cc9
Author: Patrick Bernaud <b-patrick@wanadoo.fr>
Date:   Fri Jul 11 16:02:56 2008 +0200

    Remove tests for NULL prior to freeing with g_free().
    
    Testing if pointer on memory to free is not NULL is unnecessary
    since g_free() does the test itself.

commit 21bdf2fbbad2146234fd32a0fe04bddcaddb64f8
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Jul 6 22:21:00 2008 -0400

    Updated versions in the important man pages

commit 8479d01b6a4a819ae105659a9b9f1e6ab2d85136
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Jul 6 22:03:02 2008 -0400

    Updated all ChangeLogs by running ./update-changelogs.sh 1.0-20070526..master

commit 24a38431a0d955928639311af924b8b0acf1e0db
Author: Ales Hvezda <ahvezda@-e false.(none)>
Date:   Sun Jul 6 10:03:24 2008 -0400

    Oops needed to remove outputs/spice from configure.ac as well
    
    This is an additional fix to f3ed4237658fc27a6e60e1da0c8fc9c2af2646ed

commit f3ed4237658fc27a6e60e1da0c8fc9c2af2646ed
Author: Ales Hvezda <ahvezda@-e false.(none)>
Date:   Sun Jul 6 09:55:14 2008 -0400

    Fixed common gnetlist regression tests to pass
    
    * Removed the testing of the "spice" backend completely.  It is no
      longer supported or maintained.  Removed from backends.list and
      the outputs/spice directory
    
    * Updated return codes for the drc2 backend
    
    * Allowed --regen to update one PCBBoard golden file (singlenet.sch).
      (this test is really not pcb friendly)

commit 7ae9572f269c8b973bb4a38014e28dbc10cdcb32
Author: Ales Hvezda <ahvezda@-e false.(none)>
Date:   Sun Jul 6 08:38:24 2008 -0400

    Bumped versions to 1.5.0/20080706/34:0:0 for the next development release

commit aaa1abbcfac8e67901c5b2a279e0a13f88d774b0
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat May 31 18:04:40 2008 +0100

    Tolerated changes in golden files for the vipec netlist backend.
    
    The way slotted parts are handled has changed since 1.2.0, as
    has the output of the vipec backend for slotted parts. Since
    this backend isn't likely to care about slotted parts, we allow
    these output changes as harmless.

commit ae9d82b1fdcd14acde5cbf4d54718ef3566aec80
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat May 31 18:03:50 2008 +0100

    Desired changes in golden file output for spice-sdb
    
    Slotted parts are now handled correctly. Was partly broken in 1.2.0.

commit 850a80faa3038f688ef3630aff61a05c64904abc
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat May 31 17:48:57 2008 +0100

    Update pads backend golden files to match the new \r\n line ends.
    
    The \r\n line end change was made after 1.2.0, so we need to update.

commit 345cd09f3517380a466a3254b1f6dc16ef2ce896
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat May 31 17:47:49 2008 +0100

    Remove the spice-sdb backend specific directory of gnetlist tests.
    
    They are now covered for all backends in the "common" tests directory.

commit 8a0e4e649e0aab8820fcca32214f99322788bc31
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat May 31 17:43:08 2008 +0100

    Commit golden files from netlist backends as produced by gEDA 1.2.0
    
    We want to track changes in netlist backend behaviour since version 1.2.0,
    as changes in slotted part handling since then have caused unintended
    differences in outputs for some backends. We can explicitly confirm
    thoses changes as desired or tolerable by checking in new golden files.

commit dc5a2e6b352ac4e5833ec86c0685da747362e04a
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Feb 13 16:47:40 2008 +0000

    Add a common set of tests to gnetlist evolved from the spice-sdb tests.
    
    The tests are applied to all backends, with control files being used to
    document the expected return codes, and golden files with the expected
    output. These files are generated from a 1.5.0 development release of
    of the gEDA suite, and do not in all cases represent "correct" behaviour,
    merely "gEDA 1.5.0" behaviour.

commit c1b1d3646043f27302b36a4b541bc9b34fbfe6b9
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Mar 22 19:31:04 2008 +0000

    gnetlist: Remove first-val-or-#f
    
    The first-val-or-#f function should have been removed in commit
    406234a95a4a4a6ff65c60eb2d63e32c80623a2a when its use was optimised
    away.

commit af0d9f2e8825f7d23c691deec0e70f1eaaac0e38
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Wed Mar 5 05:27:41 2008 -0500

    Updated version to 1.5.0 to mark the start of the next development cycle

commit b62e01773459c0c5df66482aa71c623191242bfd
Merge: 2915020... d68fe13...
Author: Dan McMahill <dan@mcmahill.net>
Date:   Wed Feb 27 12:40:21 2008 +0000

    Merge branch 'master' of git+ssh://dan@git.gpleda.org/home/git/gaf

commit 291502025248298b4d42cfbd80ff605dc62c3f92
Author: Dan McMahill <dan@mcmahill.net>
Date:   Wed Feb 27 12:36:12 2008 +0000

    Make the pads netlist backend use CR/LF instead of just LF.
    
    Based on a discussion on the gEDA mailing lists it seems that PADS
    (which runs only on windows) needs DOS style netlists and not unix
    style netlists.  So rather than forcing a unix2dos step in beween
    netlisting and using the result, go ahead and teach the pads netlist
    backend to directly write out a DOS style netlist.  Update the
    testsuite accordingly.

commit db168efbbe2656ef2103852e92e02ec3ce58e300
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Feb 13 14:59:45 2008 +0000

    Fix verilog and VHDL netlist backends
    
    These were inadvertently broken when making core gnetlist changes to
    accomodate slotting in spice-sdb. Since the netlist backend is loaded
    before gnetlist has traversed the schematic, the backends must not
    execute any code which queries gnetlist on load. All such work is
    to be done only when gnetlist invokes the (<backend>) method.

commit b202ee510fb90cedd038ce5ec41cc6de453dbeea
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Jan 27 11:47:13 2008 -0500

    Updated ChangeLogs for the 1.4.0 release

commit 3d43c5d642ebe12a69380e8accf5533d338ee5e5
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Jan 26 20:07:45 2008 -0500

    Bumped versions to 1.4.0/20080127/33:0:0 for the next stable release

commit ecda6ee9e101563f7fed092150532256fd9b343e
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Jan 26 19:50:16 2008 -0500

    Updated all dates and versions in all man pages

commit d2047bc2a06e6534ff32464d36f8a95aec11b9d7
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Fri Jan 11 01:11:12 2008 -0500

    Updated all ChangeLogs for this release (1.3.1)

commit 0dfb874e72040b74679152d75ab2e50b735bdbd4
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Jan 11 02:01:58 2008 +0000

    Revert "gnetlist: Strip trailing lower case suffixes from PCB package names"
    
    Stripping lower case refdes suffixes when netlisting for PCB (which
    ignores lowercase refdes suffixes) caused breakage when relying on
    matching a pinlabel=xxx to refdes=xxx inside a hierarchical
    schematic, where "xxx" had a lowercase suffix.
    
    This reverts commit e8b3be0fbcbf7518b05b437e13eeaa11478345aa.

commit c1e3075cb263a6098d2351440b680247456ae0d0
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Thu Jan 10 00:27:16 2008 -0500

    Updated and correct the gEDA/gaf man pages a little bit.
    
    This is the fix for SF Bug: [ 1556064 ] man page version out of date

commit e8b3be0fbcbf7518b05b437e13eeaa11478345aa
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 10 01:52:45 2008 +0000

    gnetlist: Strip trailing lower case suffixes from PCB package names
    
    Adds a custom (get-uref ...) function to the PCB, pcbpins and gsch2pcb
    back-ends. These strip lowercase suffixes from the uref, to ensure that
    slotted package names like IC1a and IC1b are treated as the same, IC1
    by gsch2pcb.
    
    This suffix stripping matches the fact that PCB ignores the lower-case
    suffix on net names. It would have been possible to leave the suffixes
    in the PCB and pcbpins output, although in the PCB netlist case, this
    causes any implicit power nets in each slot to be duplicated in the
    netlist, which PCB reports as an error.

commit 24431dfb4348c1fa012acb4992e3dd24a78359b5
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 10 01:52:32 2008 +0000

    Fix spice-sdb slotting to work without modified pinseq attributes
    
    Removes all previous code which iterated over slots for a given package,
    replacing it with a custom (get-uref ...) procedure which ensures each
    slot of a component is given a unique uref when gnetlist traverses the
    schematic. This means each slot will appear as a separate "package" in
    the gnetlist data-structures, even if each slot uses the same refdes=
    attribute.
    
    The format for the package uref is "(refdes/uref).(slot)", in keeping
    with the existing spice-sdb slotting code.

commit 406234a95a4a4a6ff65c60eb2d63e32c80623a2a
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Thu Jan 10 01:52:08 2008 +0000

    More efficient searching of attributes in (gnetlist:get-uref ...)
    
    Only evaluates get-attrib-value-by-attrib-name once per call, and has the
    added advantage of only defining helper functions in local namespace.

commit 3279e37ff18bfa353ada00f63d0ccf94975daf56
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 10 01:29:58 2008 +0000

    gnetlist: Add scheme procedure get-uref to determine uref of an OBJECT
    
    Moves the logic to determine an uref based on the component attributes
    into the Scheme procedure (get-uref ...), which is passed an OBJECT smob
    for interrogation.

commit fa9779d19f1a6a626948ec996735978941883f5d
Author: Bernd Jendrissek <bernd.jendrissek@gmail.com>
Date:   Thu Jan 10 00:58:35 2008 +0200

    Fix two use after free bugs.
    
    The gnetlist testsuite would sometimes screw up the terminal by
    outputting angry control characters.  This was because a warning
    message used a string that had already been freed.

commit aafa53bb3e4b248e178a654883db2ba1177c5750
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Jan 10 00:25:20 2008 +0000

    gnetlist: Load gnetlist.scm and backend before schematic traversal.
    
    Allows gnetlist.scm and backends to define scheme procedures and
    variables which may be used to control the schematic traversal process.
    
    Since some portions of of gnetlist.scm are evaluated immediately upon
    load, yet require execution _after_  schematic traversal, these portions
    have been split out from gnetlist.scm into a new file, gnetlist-post.scm
    which is loaded after traversal.
    
    The moved definitions are those of "packages", "all-unique-nets",
    "all-nets", "all-pins", and the refdes aliasing functionality.

commit 82dd20fe6d7a5022cebcc40c2a6ad73af230901c
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Jan 9 23:00:55 2008 +0000

    Change some tabs in gnetlist.c to spaces. Alter some other whitespace.

commit c29a844384a176d2880ee47aac8999ef017e50a2
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Jan 9 22:08:05 2008 +0000

    gnetlist: Clean up some old comments and EOL whitespece in gnetlist.c

commit b2380c0f70344117be7045a5fc1483698fdf4358
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Jan 5 20:57:51 2008 -0500

    Bumped versions to 1.3.1/20080110/32:0:0 for the next development snapshot
    
    Note, libgeda's so did not change at this time.

commit 5f3cde4f20975275cb38b52cd3155fda41201209
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Tue Jan 1 17:24:18 2008 +0000

    Fix configure checks for Guile (again).
    
    The previous patch missed changing the gschem configure script
    (generated from configure.ac.in), and got the logic wrong anyway.

commit b9ffb4675fbca6e92a4f713c28486e74dcbbeef4
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Tue Jan 1 15:42:13 2008 +0000

    Fix configure checks for Guile.
    
    Quote variables to protect them and alter version test logic so that
    it doesn't think version 2.1 is older than 1.8.

commit 2f7e3ae60480743a8268d392272d05bd77dd75c8
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Mon Dec 31 20:43:08 2007 -0500

    Updated all ChangeLogs for the past development release (1.3.0)
    
    This step was supposed to happen before the 1.3.0 tarballs were released.
    These changes will be picked up in a subsequent release.

commit ba474af87840a0d49e04a00de96b7df2944c09bd
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 29 09:57:43 2007 -0500

    Bumped versions to 1.3.0/20071229/32:0:0 for the next development snapshot

commit 9d3166d44b07da85ac1c4e0185daf0c67e7096ea
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Dec 29 09:52:29 2007 -0500

    Commented out test which now no longer runs correctly because of slotting fix.
    
    The slotting fix is commit 5852e531717ba947b7acca39e50f71edd5921f06.

commit e5f31fa0ba725c4a8a7dcacb47166a1468a357b6
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sun Dec 23 21:26:02 2007 +0000

    gnetlist: Make drc2 backend not emit exceptions.
    
    The drc2 backend reports drc violations by emitting an error, but it
    can do it more gracefully using the Guile display function.

commit e4c80c8a9930ebef25427053d33c5038feed31d0
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sun Dec 23 21:12:49 2007 +0000

    gnetlist: Fix segfault on failing to load schematic.
    
    Segfault was introduced when schematic load error messages were added
    in commit b635ebb641383360e5e7864830d42d13cf7f4d12.

commit fc7b46cce236e52af5d9f9b4f38061595ed5129e
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sun Dec 23 20:57:33 2007 +0000

    gnetlist: Hide a debug message from spice-sdb [1856326]
    
    A user complained that it appeared to be an error message and was
    offputting for users. Changed it to use debug-spew.

commit 3b040cf8d60a8fbe45f6e7c5f3d71a3dea119357
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Dec 19 17:48:57 2007 +0000

    Add intltoolize to the build infrastructure
    
    Provide intltoolize infrastructure for translating .desktop and MIME
    XML files. Run gettextize after intltoolize if required, so that the
    infrastructure for providing libintl on a non GNU system still works.

commit b635ebb641383360e5e7864830d42d13cf7f4d12
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Fri Dec 7 21:48:17 2007 +0000

    Print useful messages when f_open() fails.
    
    This reinstates error messages removed by the patch which modified
    f_open() to use GError for error reporting.

commit be301530f468e4d0057b3be68e218ac444c2d2be
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Fri Dec 7 17:20:55 2007 +0000

    Make libgeda's f_open() & f_open_flags use GError.
    
    This patch actually *removes* error message output from file loading
    errors is some cases.  This is for the sake of keeping the patch
    concise, and will be addressed in later patches.

commit ae225157603f063747709fa3d639ed8cdf63f6b1
Merge: 1abfd25... 43b5713...
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Tue Nov 27 07:53:37 2007 +0000

    Merge with Cesar's mingw/pu branch
    
    Merge commit 'cesar/mingw/pu' into premerge

commit 8b490a404e682ac4f1e4cbd46e22c92f58b2ae76
Author: Tomaž Šolc <tomaz.solc@tablix.org>
Date:   Fri Nov 23 21:42:50 2007 +0000

    Compilation fixes for Mac OS X 10.5 [1832417]
    
    This patch contains changes to gEDA source tree that were necessary to
    compile on Mac OS X (10.5 Leopard, PPC).

commit 43b571362e5fb663115ee1630b94c88cb74f0c13
Author: Cesar Strauss <cestrauss@gmail.com>
Date:   Sun Nov 11 18:24:10 2007 -0200

    Replace index and rindex by strchr and strrchr.
    
    These are functionally equivalent, and more portable to MinGW.

commit bdd7ef7ee9047c3942b43c165396ebe48d31e95a
Author: Cesar Strauss <cestrauss@gmail.com>
Date:   Sun Nov 11 14:17:21 2007 -0200

    Use g_path_is_absolute from GLIB instead of platform specific code.

commit 40c0582e39d6a7ed0a41e0e5deedc0bfd03bbddc
Author: Cesar Strauss <cestrauss@gmail.com>
Date:   Thu Nov 15 19:11:12 2007 -0200

    Use forward slashes as path separators on MinGW.
    
    They are also acceptable as path separators on MinGW.

commit d715ba7dc3e02d77e671b9ab54b994379daefec5
Author: Cesar Strauss <cestrauss@gmail.com>
Date:   Thu Nov 15 13:34:40 2007 -0200

    Tell git to ignore .exe files.

commit 23ead18086e59aab22945c37ff95f49e5747f277
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Oct 17 04:50:39 2007 +0100

    Remove the ATTRIB linked list in favour of a GList
    
    The ATTRIB.copied_to pointer is moved from the ATTRIB struct to reside in
    the source OBJECT being copied.
    
    The ATTRIB *attached_to pointer of an attribute's text OBJECT is changed
    to OBJECT *attached_to, pointing directly at the OBJECT to which that
    attribute is attached. This avoids the need to traverse the ATTRIB list
    and retrieve this pointer from the HEAD attribute node. (A HEAD node is
    not used in the GList of ATTRIBs).
    
    The ATTRIB structure now only contains one element, OBJECT *object. This
    points to the text OBJECT holding and displaying that attribute's
    "name=value".

commit 4595948e41b2d195eb78a820fc15d0d8596067a4
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Oct 4 02:34:50 2007 +0100

    Avoid accessing (GList)->{next|priv} directly to aid debugging.
    
    Instead use g_list_{next|previous}, allowing us to catch errors where
    changing from gEDA structures which had their own next and prev pointers.
    
    To debug, edit the glist.h include file from glib, renaming the next and
    prev elements in the GList structure and macros. Recompile gEDA to see any
    bad accesses such as those where the HEAD nodes were previously skipped.

commit 2e3c06cbcdce8254f70162ecebd650980cf70ee9
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Sep 9 00:18:34 2007 +0100

    Remove the globally exported variable_set_func.
    
    Only gattrib provided a non-NULL export of this symbol, and such imports
    into libgeda prevent libgeda being dynamically linked on some platforms.
    
    To better match the other tools in the suite, gattrib's i_vars_set() and
    i_window_vars_set() functions were combined into a single i_vars_set().

commit 8952ec9bd4f4985fae8aadd6c4103b933464152a
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Sep 9 00:01:14 2007 +0100

    Remove unused extern variables from libgeda, gnetlist and utils.

commit adca8b7d5897b11331c6a20392d22ad1e58ee572
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Mon Sep 3 12:15:43 2007 +0100

    Replace the linked list of PAGEs with a GedaList (typecast to GedaPageList)

commit 086908ed44c9abc6d5907888695933b59ead4684
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Sep 2 22:48:59 2007 -0400

    Last minute updates to the READMEs for all projects
    
    Cherry picked from stable-1.2 branch, since these changes apply to master
    too.
    (cherry picked from commit 88fcb8d1bcdaa1436572d6a0ebe86f4df242a8d6)

commit 0aad767df307a70d9fefc73fb08a49908cc26706
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Sep 2 16:50:29 2007 -0400

    Fix for Bug#1776547: graphical versus source
    
    The usage described in Bug#1776547 is pretty unique, but it might be handy
    to support in general.  The recursive source= and graphical= behavior is
    a unexpected side effect of the traversal of graphical symbols/components
    (552e06e921e4b6eb758b9483d1913c39b3fef5f5).  The fix is to explicitly
    check for the graphical= attribute and if it is found to not further
    traverse the hierarchy.  Regression tests were added so this behavior
    doesn't change in the future.

commit 6056ef48272778419d39e0233f395bfd60019ab9
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Aug 19 22:40:19 2007 -0400

    Update ChangeLogs using: ./update-changelogs.sh 1.0-20070526..master

commit 92b3b65d0cec286b0ce59fc43cc07dcb60c39fa1
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Aug 18 16:14:30 2007 -0400

    Bumped versions to 1.1.2/20070818/30:2:0 for the next development snapshot

commit 72c4db8aa20c22baa4243a0d27b3bb3f32f3ef64
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Tue Aug 14 07:46:06 2007 -0400

    Moved 3 rc promotion keywords from gschem into libgeda (fix for bug#1748143)
    
    All the promotion keywords were moved from gschem into libgeda to fix:
    [ 1748143 ] gschlas causes duplicate prompted attributes   The primary
    cause of this bug was that the promotion variables within TOPLEVEL were
    not being setup by gschlas and so duplicate refdes attributes were
    showing up in the schematic.  The three rc keywords that were moved are:
    
    (attribute-promotion ... )
    (promote-invisible ... )
    (keep-invisible ... )

commit 703151a94b0eece6b24b3b18ce2465eeae068814
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Aug 9 12:36:17 2007 +0100

    Remove old prototypes from various prototype.h where no function exists.

commit 71600215311af0a98ed371ab750b09182de139a1
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Aug 11 00:52:25 2007 +0100

    Revert inadvertent commits:
    
      dbfe00a20de3f875e519ef60629b968977fc5db8 (commit)
      f1451161914a0ed0bb43123279906d82cb7d5b49 (commit)
      2ad315fc0badd16a9614d1107b22c973908893d5 (commit)
      d8844bb48969aa45a05cbe6836fc96c198e88c19 (commit)
      ee1b64a1ae07eb99eae5d63307e9fe23f9604662 (commit)

commit d8844bb48969aa45a05cbe6836fc96c198e88c19
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Fri Aug 10 21:26:10 2007 +0100

    Replace the linked list of PAGEs with a GedaList (typecast to GedaPageList)

commit 0f5c72daacfa7d78666769e4063122934861da5c
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sun Aug 5 22:36:56 2007 +0100

    Alter guile 1.6 compat tests to work with older autoconf versions
    
    Don't use AC_DEFINE to define function like macros, instead use
    AC_CHECK_DECLS to set a HAVE_DECL_..., then use AH_VERBATIM to
    insert a test which defines a replacement if necessary.

commit 5c098070b87709540345afb1b3b79d9db470ed53
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Thu Aug 2 22:09:52 2007 +0100

    Fix configure checks enabling guile 1.6.x compatability routines.

commit c31bd4adb677d671af8bdf5aed71f96a4995f9b1
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Tue Jul 31 20:03:51 2007 +0100

    Changed configure.ac(.in) to allow guile-1.6
    
    Introduces compatibility #defines where guile 1.8 funcs are missing in
    guile 1.6. Also adds missing configure.ac checks for guile in packages
    gattrib, gnetlist, gsymcheck and utils.

commit 525b4a0ff3a1b8cc321f117513796e4571702da8
Author: Bernd Jendrisse <bernd.jendrissek@gmail.com>
Date:   Sun Jul 29 09:56:33 2007 -0400

    Applied patch to use $(GROFF) instead of which groff

commit 31f65059cf1ce5be9b8a0987d4538fed075a6805
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Wed Jul 25 02:09:15 2007 +0100

    Fixed two calls to free() where g_free() should have been used.
    
    Updated Doxygen comments to note where memory should be freed with g_free()
    rather than free(). Commented three instances where free() is correct.

commit e978b9098c7567ad3b9c834a5e4914de3bc039ec
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Jul 8 18:18:39 2007 -0400

    Updated ChangeLogs using ./update-changelogs.sh 1.0-20070526..master

commit d7e8f5abc35644e90bfabf84396023cab09f2968
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sun Jul 8 16:18:57 2007 -0400

    Updated configure.ac[.in] and Makefile versions in prep for a release.
    
    The update was done by running ./update-versions.sh 1.1.1 20070708 30:1
    at the toplevel.  libgeda's shared library version is now 30:1:0

commit 3f3505662fe4bb8c585aea0ea965296bd923956c
Author: Carlos Nieves Onega <cnieves@iespana.es>
Date:   Sun Jul 8 21:16:17 2007 +0200

    Added *~ to .gitignore files.

commit 0cde8c1cf6c863946c3d4acdc9e9165c99526bef
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Thu Jul 5 01:52:04 2007 -0400

    Updated gnetlist ChangedLog to include the new backend addition

commit b692c010f9ccf0b89ff1ad653ff1c5a812887cc3
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Thu Jul 5 01:43:47 2007 -0400

    Added new gnetlist backend by John P. Doty

commit dd995d808b9f4440db48d2daa2473a9d9cfd88ef
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Thu Jul 5 01:36:31 2007 -0400

    Updated ChangeLogs using ./update-changelogs.sh 1.0-20070526..master

commit 099c49e5689afda4b8a5221dde73f9fdf1972ce7
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Thu Jul 5 00:11:45 2007 -0400

    Updated all version strings to 1.1.0 and 20070705

commit 87731dda219b0986fed0d1a7789e17678dd79d8c
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Wed Jul 4 23:55:07 2007 -0400

    Removed AM_MAINTAINER_MODE from all configure.ac[.in] files.
    
    Maintainer mode is no longer needed or desired now that we are no longer
    using CVS.  Hopefully this will prevent some of the odd problems that were
    observed during the last release when the Makefile's didn't get auto
    regenerated.

commit dc7b8c53f4bc94626634887e46167a367a91f707
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Wed Jul 4 23:32:15 2007 -0400

    Changed gEDA/gaf's version mechanism to use new dotted X.Y.Z convention
    
    Added DOTTED_VERSION and DATE_VERSION variables to all configure.ac[.in]
    files.  Changed all version related output to be DOTTED_VERSION.DATE_VERSION.
    Updated the configure.ac[.in] files to have consistent top sections.
    Updated code and various scheme files to use the new DOTTED_VERSION
    and DATE_VERSION variables available via config.h.  Stopped using
    PREPEND_VERSION_STRING as the mechanism for doing the dotted versions
    (not intended for that use).  Updated update-versions.sh script to set the
    version in all the configure.ac[.in] files and the toplevel Makefile.
    Went through all the g_rc.c files and make the handling of rc file
    mismatch handling consistent.

commit 5237ea395af0ac1c4cecd18ea2a28291e3d48a90
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Wed Jul 4 23:22:14 2007 -0400

    Updated spice-sdb golden file to pass make check.

commit b44a1e527497e55471a75b0af410e78310ed50d8
Author: Dan McMahill <dan@mcmahill.net>
Date:   Tue Jun 26 18:11:39 2007 -0400

    guile-1.8.1 fixes and testsuite additions.
    
    Updated the cascade and pcbpins backends to work with guile-1.8.1.
    While here added a testsuite entry for these two backends.

commit f8680b18c8e27617eccbbf158acb84ccac7ac386
Author: Dan McMahill <dan@mcmahill.net>
Date:   Tue Jun 26 16:20:58 2007 -0400

    Portability fixes for the testsuite.
    
    Get the testsuite fully working on SunOS-5.9/sparc.  This
    entails two issues.  The "-I" option to diff is a GNU
    extension and is not supported by the solaris diff.  Work
    around this by using sed to filter out the lines which were
    to be ignored.  This addresses the tests run by gnetlist/tests/runtest.sh.
    
    The other issue is that the longopts tests in the spice-sdb
    testsuite should not be run on a system that does not have
    getopt_long().  To address this, extend the run_tests.sh to allow
    a variable in the tests.list file to disable a test unless set to "yes".

commit 57b2848a7e69d19e18ac03a1264cb55d41c65d0e
Author: Ales Hvezda <ahvezda@laptop.ales.net>
Date:   Mon Jun 25 22:31:07 2007 -0400

    Changed/updated the #define mechanism for creating custom gEDA/gaf versions.
    
    Previously this mechanism was used to create custom versions of gEDA/gaf.
    With the coming of stable and development versions of gEDA/gaf, this
    mechanism was changed so that end users can tell the difference between
    a stable release vs a development release (by looking at the version string).
    The CUSTOM_VERSION #define was changed PREPEND_VERSION_STRING and is now
    prepended to any strings that output the VERSION string.

commit dc0917b142d9c00af33065ce9eaac8b0a6dc3448
Merge: fee96ef... 0ad58a2...
Author: Dan McMahill <dan@mcmahill.net>
Date:   Mon Jun 25 09:45:22 2007 -0400

    Merge branch 'master' of git+ssh://dan@git.gpleda.org/home/git/gaf

commit fee96effc220e78acddf9721a6bd9d0c34bef881
Author: Dan McMahill <dan@mcmahill.net>
Date:   Mon Jun 25 09:39:17 2007 -0400

    Add missing reference files to distribution and re-enable checks.
    Updated the list of reference files that go into EXTRA_DIST so that
    'make distcheck' can work again.  Now the this is fixed, re-enable
    the spice-sdb teststuite.

commit 0ed9cf080cfc8509cb29cbe4e976f86b2821c407
Author: Ales Hvezda <ahvezda@laptop.ales.net>
Date:   Sun Jun 24 23:48:51 2007 -0400

    Updated copyright text/header in most source files.
    
    Updated all copyright text dates that have my name (Ales) associated
    with them.  Added gEDA contributer copyright text where appropriate.
    Fixed a bunch of GNU text typos (these should have been either GPL
    or gEDA).  Removed some $Id$ tags which are no longer used/needed.
    Fixed a few file header that were just plain incorrect.

commit a4e956c6272ea982ec54327dcc5e3aa528ac8c99
Author: Ales Hvezda <ahvezda@laptop.ales.net>
Date:   Sun Jun 24 22:22:32 2007 -0400

    Added *.o (all object files) to all src .gitignore files.

commit edff766e8b15c311505810241528026f722e30b6
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Jun 23 18:19:11 2007 +0100

    Make sure various files are installed.
    
    Several new files were not being distributed. This patch addresses the
    problem by adding files to EXTRA_DIST in automake input files.

commit e57c37a66edc1c4d4e4d7160001749df51b96c4b
Author: Peter Clifton <pcjc2@cam.ac.uk>
Date:   Sat Jun 23 18:17:29 2007 +0100

    Move all .cvsignore files to .gitignore files

commit 065595ccd8b5e77616c29a37742357cfab0ae7f7
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Sat Jun 23 16:11:44 2007 +0100

    Changes to ChangeLog usage.
    
    With actual changes this time.  Old-style changelogs as of gaf 1.0
    (release of 26th June 2007) are moved to ChangeLog-1.0; new (blank)
    ChangeLogs replace them.  A script has been added to the gaf top level
    to allow maintainers to update all the ChangeLog files immediately
    prior to making out a release.

commit b7089bcc123e52bbadd6e635c823d937e620b75b
Author: Ales Hvezda <ahvezda@seul.org>
Date:   Sat Jun 9 18:29:22 2007 +0000

    Commited new version of gnet-calay from John with various fixes.

commit eed0be648afd9bc421a8d0059de621df778c5fbe
Author: Carlos Nieves Onega <cnieves@iespana.es>
Date:   Sat Jun 9 16:09:36 2007 +0000

    Avoid writting '\r' to files on MinGW.
    
    Applied patch #1733826 from Cesar Strauss. Thanks.
    When using the "w" mode in fopen, MinGW translates LF into CR-LF.
    This creates havoc when reading the same file again, when sharing
    schematics files with other systems, and when placing them in a VCS.
    
    This can be avoided by using the "wb" mode. The "b" is ignored on
    all POSIX conforming systems, according to these sources:
    
    http://www.die.net/doc/linux/man/man3/fopen.3.html
    http://www.opengroup.org/onlinepubs/009695399/functions/fopen.html
    
    There are other ways, however. Adding
    
    #include <fcntl.h>
    int _fmode = _O_BINARY;
    
    on some source file should work as well. Another way is linking
    against the MinGW supplied binmode.o, which does exactly that.

commit 3d58ffc27e6d915c4b1ebbd6f6b96d98d6126b7b
Author: Peter TB Brett <peter@peter-b.co.uk>
Date:   Mon Jun 4 16:15:37 2007 +0000

    Replace deprecated Guile functions & macros.
    
    Replace deprecated SCM_STRINGP, SCM_INUM, SXM_INUMP & SCM_MAKINUM with
    recommended equivalents for Guile 1.8.


#GCL#