I'm glad to see 0.15 is out! There's been a lot of work put into this, and thank you for heading it all up Chris!
Christopher D. Stover -------------------------------------------------- From: "Chris Frey" <cdf...@foursquare.net> Sent: Saturday, August 29, 2009 2:48 AM To: <barry-annou...@lists.sourceforge.net>; <barry-devel@lists.sourceforge.net> Subject: [Barry-devel] Barry version 0.15 released > The Barry development team is pleased to announce the release of > version 0.15. Source and binary packages are available for immediate > download at the URLs below. > > Barry is an Open Source application that provides synchronization, backup, > restore, and program management for BlackBerry devices. Barry is primarily > developed on Linux, but is intended as a cross platform library and > application set, targeting Linux, BSD, 32/64bit, and big/little endian > systems. > > This is a long awaited release, with lots of new features from 11 months > of development: > > - binary packages for Ubuntu 9.04 and Fedora 11, and better support > for Fedora 11's HAL/udev system > - a new version of the opensync plugin in step with opensync's 0.3x > devel > tree, which includes new syncing options for tasks and memos > and photos (this plugin is currently considered experimental > until opensync officially releases 0.40) > - improved sync logic for contacts (both 0.22 and 0.4x plugins) > - a FUSE filesystem driver to browse BlackBerry databases as files > - internationalization improvements in library and sync > - support for the BlackBerry Storm and Pearl Flip > - javaloader support (includes load, save, erase, info, eventlog, > screenshot, wipe, settime, etc) > - fixes to support newer kernel USB behaviour changes > - zsh command line completion support > - parsing support for RIM's USB logs > - database parsing improvements: phone call logs, SMS records > - new modem chat scripts: China Mobile, AT&T Cingular > - backup GUI can now pick backup destination folder > - backup GUI can now change between multiple devices > > > > Binary packages are available on SourceForge for 32 bit systems, for > the following distros: > > Debian stable (lenny) > Ubuntu 7.10, 8.04, 9.04 > Fedora 7, 8, 9, 11 > > Source packages are available in Debian, RPM, and tarball formats. > > http://sourceforge.net/projects/barry/files/ > > Packages used to be available on the OpenSUSE Build Service, but > this has been discontinued for this release due to lack of time. > > Detailed user documentation is available online at the usual place: > > http://www.netdirect.ca/software/packages/barry/ > > (scroll down to the 0.15 section, where the files are > grouped more clearly that in the "new files" list) > > There are some known issues in this release (which affect older > releases as well). Namely: > > 1) Restoring backups for some databases on newer Blackberries > doesn't work (for example, on the 8120, 8700g). Before > relying on barrybackup, please make full data backups using > RIM's Windows tool. > > 2) Syncing is not supported on Fedora Core 9 and 10, since they > packaged the OpenSync 0.3x devel tree. This has been fixed > in Fedora 11, as they returned to 0.22. > > 3) Accessing the database (such as during a backup) while copying > files using the usb_storage kernel module may cause some > Blackberries to spontaneously reboot. > > For full details regarding the changes in 0.15, please see the ChangeLog. > > For information on release target dates and milestones, see the > latest TODO file in the git repo: > > http://repo.or.cz/w/barry.git?a=blob;f=TODO;hb=HEAD > > You are encouraged to join the project, play with the code, and report > any problems or feature requests to the mailing list. You can use the > command line tool to capture unsupported database items and help in the > reverse engineering process. > > Enjoy! > > > Contributors in this release: (alphabetical order by last name) > Richard Esplin - AT&T Cingular chat script > Ron Gage - Autotext and Service Book parsing notes > Dr. J A Gow - iCal sync and category patches > Andy Herkey - modemtest.rb script > - ipmodem password support for Bold > Josh Kropf - many additions to javaloader support > - support for large java modules > - brimtrans tool for parsing RIM USB logs > John Ladan - Pearl Flip ID report > Ryan Li - SMS record parsing support > - China Mobile chat script > - backup GUI improvements > Martin Owens - HAL FDI scripts > - devel snapshot binary packages for Ubuntu > - documentation of Content Store > Bill Paul - FreeBSD and build system debugging > Rick Scott - reverse engineering infomation on various > details in the javaloader protocol > Jose Sogo - Debian package patches > Christopher Stover - Fedora 10 debugging help > Nicolas Vivien - BlackBerry Storm support > - initial javaloader patch > - porting sync plugin to opensync 0.4x > - zsh completion script > - sync support for photos > - sync support for memos and tasks > - phone call log parser > - bookmark documentation > - build system fixes (boost) > - LDIF patches > Robert Yaklin - VSM file format documentation > > A special thanks to all testers who have provided valuable feedback. > > Thanks also to the official distro maintainers: > > Debian Jose Carlos Garcia Sogo > Fedora Christopher D. Stover > Mandriva Adam Williamson > > Thanks to Christopher Stover and Martin Owens who provide development > snapshot binary packages for Fedora and Ubuntu, respectively. > > > > > ChangeLog for version 0.15: > > Release: version 0.15 - 2009/08/28 > ------------------------------------------------------------------------------ > 2009/08/28 > - added binary build for Ubuntu 9.04 to maintainer scripts > 2009/08/27 > - added binary build for Fedora 11 to maintainer scripts > 2009/08/21 > - updated www documentation for upcoming release > - applied Nicolas Vivien's phone duplication patch (opensync 0.4x) > - fixed missing conflict check in vcard code in opensync 0.4x > plugin > - backported new TYPE parsing code from opensync-0.4x plugin to > old opensync 0.22 plugin > - added X-EVOLUTION-UI-SLOT support in both sync plugins to retain > phone number order > 2009/08/17 > - added pin and password command line support to bfuse > - updated bfuse man page > 2009/07/29 > - added border to barrybackup doc images > 2009/07/28 > - updated license and URL in barry.spec > - updated barrybackup Credits dialog to reference AUTHORS file > - updated barrybackup.1 man page > - updated doc/www scripts to generate more dynamic pages, > so subheaders are not hard coded... this helps support > updating the new NetDirect website > - commented out out-of-date docs in hacking.php > - marked milestone 4 complete (javaloader) > - linked device compatibility list from index.php > 2009/07/10 > - changed pppob to report if Serial was fallback or command line > - started device compatibility list in web docs > 2009/07/07 > - applied Ryan Li's GUI enhancement patches, which let the user > switch devices without restarting, with fixes: > - reorganized BackupWindow.glade for better spacing > - fixed width issue (main window was set to 275 > width, which made things too narrow on Debian > Lenny > - moved buttons beside the drop down box, to > mimic old GUI behaviour > - added space around the middle controls, but > not around the menu bar or status bar, which > should be right against the outer window. > - renamed m_device_num to m_device_count for clarity > - changed m_device_count to unsigned int to fix compiler > warnings > - removed gtkmm call to unset_model() > The function unset_model() is only available in gtkmm > 2.16, and Debian Lenny only has 2.12, and won't compile. > Since set_model() automatically does the unset for us, > just reorganize the calls so Lenny is supported. > 2009/07/06 > - applied Ryan Li's SMS clarification patch > 2009/07/02 > - added Martin Owens' new Ubuntu PPA link to docs > - added DEPUTY documentation and updated AUTHORS > 2009/06/30 > - documentation updates > - added SMS to list of parsers in documentation > - updated sync notes for Fedora 9 and 10 in documentation > - removed Martin Owens' PPA binary package link, since > I can't find the binaries anymore > - reverted commit 2d0e60d1: put hal fdi files back > - updated HAL FDI files for Fedora 11 > - duplicated directory hierarchy under hal > - added 19-blackberry-acl.fdi to add the access_control > capability to blackberry devices when they are > plugged in. As long as this file is under > the 10osvendor directory, then > 20-acl-management.fdi runs after it, adding > the hal-acl-tool callout, which does the low > level getfacl/setfacl work on the /dev/bus/usb/*/* > device file. > related thread: > > http://lists.freedesktop.org/archives/devkit-devel/2009-June/000247.html > - changed hal fdi policy to use "pda" instead of "scanner" > - applied Ryan Li's 7-bit GSM conversion patch for SMS records > - applied Ryan Li's barrybackup patch: can now select backup path > 2009/06/26 > - fixed install location of hal fdi files in debian and rpm > HAL reserves /usr/share/hal/fdi/information/10freedesktop > for files that the project itself distributes, so > we get to use 20thirdparty. > 2009/06/24 > - cleaning up autoconf warnings on Fedora 11 > - changed buildgen.sh to run libtoolize first for all subprojects, > otherwise, a later project's run may copy files into ../m4 > after having created configure for a previous project. > This would cause a makefile discrepency during the build > on some systems (Fedora 11), and cause configure to be > incorrectly regenerated mid-build, due to timestamps. > 2009/06/21 > - updated dependencies documentation page > 2009/06/20 > - added BadPacket exception, for socket errors that need to meddle > with the response code > - added special case for JavaLoader's Goodbye packet > On older devices, such as the 7750 and 7130, they respond > with a NOT_SUPPORTED error instead of an ACK, for Goodbye > packet commands. This is seen in Windows traffic as well, > so should not be considered an error. Thanks to Josh Kropf > for confirming this protocol behaviour. > - updated socket.cc to use new BadPacket exception where > appropriate > - pulled from Josh Kropf: 0f1cca2fa69d0810d439ade4fb3dd0dfa15c89a6 > - fixed size issue with deviceinfo command on 7130 devices > 2009/06/19 > - reworked autoconf rules to handle every kind of Boost dependency: > - let user enable/disable from configure, default to > disable, but error if enabled and not available > - let user specify separate include and lib paths, since > some Boost installations have an additional > boost-1.34.1 style directory in them > - default to no path overrides, using system > defaults > - let user specify name of serialization library, since > the name of the library can change whether you're > building from source or not > - default to searching for boost_serialization or > boost_serialization-mt, and error if not > found > - New configure switches are: > --enable-boost Handles enable/disable > --with-boost-include=path Override the include > path > --with-boost-lib=path Override the lib path > --with-boost-serialization=name Override the name of > serialization library > to link with > - updated documentation, build test, and binary packages to use > new switches > 2009/06/18 > - added non-backward-compatibility note to man/btool.1 > - applied Nicolas Vivien's LDIF patches, clarifying split > between work and home addresses, and adding support > for the homePostalAddress LDAP field > - fixed missing #includes for Fedora 11. Thanks to > Nathanael Noblet for reporting this. > - applied Ryan Li's PPP chat script for China Mobile > - added barry-chinamobile chat script to RPM and Debian packages > - renamed blacklist-berry_charge to blacklist-berry_charge.conf > for Debian sid and Fedora 11. Thanks to Rodrigo Linfati > for reporting. > - applied Nicolas Vivien's autoconf patch to automatically > search for boost_serialization or boost_serialization-mt > 2009/05/12 > - applied Dr. J A Gow's category patches (task + memo) > - fixed compile error when using Boost > - cleaned up Categories handling: > - s11n-boost.h > - Dump() should be consistent across Calendar, > Memo, Task > - fixed i18n support to category names > - cleaned up code formatting to match surrounding code > - moved CategoryList to its own class, and removed duplicated > string conversion code > - removed protocol-specific MemoType and TaskType fields from > their corresponding record classes > 2009/05/09 > - applied Nicolas Vivien's Task builder patch for the Storm > and Bookmark documentation patch > 2009/05/08 > - applied most of Dr. J A Gow's recurring iCal sync patch > - did not include the pure upper case changes, since > it is an RFC bug in SynCE > see > http://www.mail-archive.com/barry-devel@lists.sourceforge.net/msg01109.html > - added starttime to RecurToBarryCal() args, to make the > prerequisites clear (StartTime must be determined before > recurrence is parsed) > - added check that COUNT is not zero (RFC requirement) > - fixed timezone drift: using gmtime() + mktime() uses two > different timezones... use localtime() + mktime() > instead > - comment resizing > - reformatted if/else FREQ tests into a more readable > sequence, and fixed a logic error in the yearly count > advance, which should only happen when FREQ == YEARLY > - updated AUTHORS > 2009/05/06 > - going through www documentation, updating for 0.15 > - updated man/btool.1 with new -a option > - updated TODO list with new reality > - applied Nicolas Vivien's phone call log parser patches > - added CallLog to s11n-boost.h > - added size checks to parser > - removed CallLogType, since it is protocol specific > - added and tidied comments > - made CallLog::Clear() match the class declaration order, > fixed missing member, and used enum names to set > enum defaults > 2009/05/05 > - applied Nicolas Vivien's task sync patches, with btool updates > - fixed "VTOTO" typo > - now that there's a Desktop::ClearDatabase() call, > use it in SaveDatabase() to reduce code duplication > - use separate vector<> clearDbNames in btool > to eliminate chance of mixing -d and -a argument > types and losing data > - reverted VNOTE change in vformat.c > - added VFORMAT_JOURNAL support to vformat.c based on > r2795 of the opensync vformat plugin > - fixed some TimeZoneValid flag mistakes in Task and Calendar > - display DueDateFlag in src/r_task.cc > - changed vtodo.cc so DTSTART is only set if available in BB data > 2009/05/01 > - applied Nicolas Vivien's zsh patch > - applied Nicolas Vivien's memo sync patch > - added MEMO_TYPE 'm' to built memo record > - updated btool to include new Memo builder > - updated config.rpath to latest from Debian stable > 2009/04/27 > - removed free() in tarfile.cc, since some distros have bugs in > th_get_pathname() that doesn't return a freeable string > 2009/04/22 > - fixed some missing headers as reported by Christopher Stover > - applied Ryan Li's SMS body null terminator strip patch > - fixed code formatting issues > - added comment on why we're stripping out nulls > - fixed potential single byte buffer overflow > - pulled from Josh Kropf: 8266c95a94 > added "reset to factory" command to bjavaloader > - split creation of deb src tarballs into separate script > instead of tar-create.sh > - added more statements for the m4/ directory in configure.ac > and Makefile.am, attempting to eliminate build warnings > on FC10, as reported by Christopher Stover > 2009/04/14 > - applied Nicolas Vivien's sscanf patch (opensync-0.4x) > - wrapped OSyncList in vSmartPtr<> to automatically handle > all calls to new osync_list_free() API > - applied Ryan Li's updated SMS parser patch > - moved SMS metadata structure into its own struct > in protostructs.h > - fixed some endian conversion errors > - added strnlen() call to avoid null terminators in > the Addresses strings > 2009/04/10 > - added brimtrans to rpm spec file and debian install > 2009/04/09 > - fixed OSyncList* leak in 0.4x plugin from call to > osync_plugin_info_get_objtype_sinks() > 2009/04/08 > - added IConvHandle, making it possible to support more than one > charset in Blackberry data > - removed patches from opensync-plugin-0.4x 'make dist' target > - updated buildtest.sh for opensync 0.4x > 2009/04/06 > - updated man/bjavaloader.1 > 2009/04/04 > - added Product ID 8001 for the Pearl Flip. Thanks to John Ladan > for reporting the new ID. > 2009/04/03 > - removed 0.22 opensync patches from 0.4x tree > - applied opensync Photo support patches from Nicolas Vivien > - renamed SequencePacket() to HideSequencePacket() for clarity > This is a temporary measure... this sequence packet > code needs to be cleaned up. > - added Photo support to examples/addcontact.cc > - removed try/catch from vcard/vevent, pending feedback from > Nicolas > - applied Ryan Li's SMS record parsing patch > - changed capitalization to match other code > - added Sms to s11n-boost.h and required operator > - added Sms to btool.cc, record.h, and Makefile.am > - added size checks to the parser code > - cleaned up brackets and casting > - made dump output prettier (at least to me) :-) > - added "unknown" state for MessageStatus, as default > - cleaned up whitespace > - updated AUTHORS > - added endpoint override to Probe class while testing the 7130 > When using btool's -e option, the Probe class still > attempted to probe endpoints as it saw fit. > 2009/04/02 > - fixed double-free bug in opensync 0.4x plugin > Can only report error or success once in a change. > - updated buildgen.sh to handle ctags for multiple opensyncs > - fixed inverted logic when doing a hashtable slowsync (opensync > 0.4x) > - fixed uninitialized osync_trace args in both vformat.c files > 2009/04/01 > - opensync 0.4x changes due to upstream API updates > - removed idmap and cache code, in favour of opensync's > uid mapping and hashtables respectively > 2009/03/31 > - opensync 0.4x changes due to upstream API updates > - removed deprecated opensync header > - updated 0.4x plugin to use new slow-sync API > - updated 0.4x plugin to use new function registration and > user data API > - updated 0.4x plugin to use new sink list (API change) > 2009/03/27 > - changed PIN format from decimal to hex in opensync-0.4x plugin > config > - added special case so photo data does not print raw in btool > dumps > - fixed call to trace.logf() in plugin code... should never call > a printf-like function unless you use a constant "%s" > - made opensync-0.4x:vformat.c's base64 routines const-correct > - fixed free()-crash bug in both vformat.c files > When generating the resulting VCARD data, vformat.c > assigned > static strings to a dynamic heap list pointer, and the > later free always failed. > - merged some changes from Jose Sogo's Debian package work > - menu changes to conform with policy > - renamed opensync plugin package name to match Debian > - added Nicolas Vivien and Josh Kropf to debian/copyright > - fixed uninitialized variable in SocketZero > - fixed memory leak in tarfile.cc > - fixed uninitialized time struct in bjavaloader.cc (valgrind > error) > 2009/03/24 > - fixed typo in zsh/_bjavaloader thanks to feedback from Nicolas V. > - pulled from Josh Kropf: 75b1a89e8b1eb9d89bcd8b5ecbed9a47af5a8930 > added logstacktraces support to bjavaloader > - updated TODO list with "- All" database parsing > 2009/03/13 > - added git tag signing to maintainer scripts > - added signed tagging to release checklist doc > 2009/03/11 > - updated opensync-plugin-0.4x/README with build help > 2009/03/10 > - pulled from Josh Kropf: d6c3173ccd87c6223647950963da79331a68d712 > added deviceinfo support to bjavaloader > - pulled from Josh Kropf: 272f371e8bb95b6c41a8293de657da45a782736f > new tool: brimtrans for dealing with RIM USB logs > - pulled from Josh Kropf: 7ed1df560ea7cc6a2205d8f8627c00c7aaed2294 > added wipe support to bjavaloader > - added confirmation step to bjavaloader wipe, so it is harder to > accidentally wipe your device > - applied zsh patch from Nicolas Vivien for bjavaloader wipe > 2009/03 > - changes inspired from Rick Scott's XmBlackberry > - event log timestamp is a time_t in milliseconds > 2009/03/07 > - added Mode base class for common mode class code > 2009/03/06 > - reorganized external links by date and topic, and added link > to Nicolas Vivien's Blackberry on Linux documentation > 2009/03/05 > - pulled from Josh Kropf: b3d3a6f48fe008e8842c057df4a653a04b9c56ce > - compile fix when using __DEBUG_MODE__ > - fixed timeout when calling JavaLoader::GetDir on device > with zero modules > 2009/03/04 > - fixed embarrassing incorrect exception messages in usbwrap.cc > Fortunately, these didn't affect my timeout testing. > - added Data::QuickZap(), so that usb read errors don't > return old data > - inlined Data::QuickZap() > - added usb_set_altinterface() call to Interface class > This should hopefully workaround the change in 2.6.28 > kernels where set_interface is not called by the kernel > automatically anymore. > 2009/03/03 > - applied Nicolas Vivien's opensync-0.4x API patch > 2009/02/28 > - applied Nicolas Vivien's zsh patches > 2009/02/26 > - added opensync-0.4x to tar-prepare.sh maintainer script > 2009/02/24 > - added note about novj option issue in barry-att_cingular > - added gettext to documentation dependency list (for iconv.m4) > 2009/02/20 > - clarified bjavaloader save behaviour > - updated bjavaloader man page > - added maintainer/build as exception for test/buildtest.sh > - merge opensync-0.4x branch into master: > 2009/02/20 > - removed nested m4 in preparation for merge into > main Barry tree > 2009/02/18 > - applied Nicolas Vivien's opensync-0.4x > Calendar+Contact patch > 2009/02/01 > - applied Nicolas Vivien's opensync-0.4x > porting update patch > - removed unneeded ChangeLog.osync > - added opensync-plugin-0.4x to top level configure, but not to > binary packages > - added old emails from Ron Gage from 2006 regarding Autotext > and Service Book parsing > 2009/02/19 > - pulled from Josh Kropf: 9b63a99867b3314b8d42212cac05a6f28860d37b > - saving large modules in pkzip format > - cleaned up StartStream function; use JLPacket instance > - refactored SendStream; use JLPacket instance and use > simple > function for seeking to next cod file in input > stream > - made following changes to Josh Kropf's patches > - Added CODFILE_TYPE_SIMPLE check back in > - Minor change of c_str() to data() where a null terminator > is not needed > - Fixed unlikely buffer overflow in m_javaloader.cc > Goal: never trust outside sources of > information (including that from the device) when > managing our own buffers. > - Removed CodFile forward declaration, since class does not > exist anymore > - Fixed typo in configure.ac, and made missing zlib warning > more emphatic > - Added clarifying comment to CodFileBuilder > - Fixed potential reference bug in CodFileBuilder > constructor > It is possible that the const std::string& passed > into the constructor is a temporary, so make a copy > of it instead of just holding a reference. > - Removed multi-unget call in cod.cc -- only one unget() is > guaranteed in C++. It appears that the unget is > unnecessary, since the following code seeks to end > and back to beginning anyway. > - Added copyright line for Josh Kropf in cod.cc and > cod-internal.h > - Fixed incorrect zlib check, since zlib doesn't support > pkg-config on most systems yet. Changed it to > use AC_CHECK_LIB, and added a --with-zlib configure > option which will cause configure to halt if > zlib is not available. Added --with-zlib to > binary package builds, and to build test script. > - added zlib to rpm and deb build requirements list > 2009/02/18 > - updated USB capture logs archive page with new logs from Josh > Kropf > - applied as patch Josh Kropf's annotated javaloader-reset.txt log > - added note about windows registry keys to USB-capture.txt > 2009/02/17 > - applied Josh Kropf's patch adding support for forced erase and > load while module in use, properly handling when the > device requests a reset > 2009/02/15 > - applied Nicolas Vivien's RecordId patch for Contact class... > this updates Barry to deal with the new record ID behaviour > on the Storm > 2009/02/13 > - removed unneeded call to .c_str() in EmailAddress stream operator > - added support for Calendar fields: Organizer, Accepted, Invited > - reorganized the udev/ script directory, based on distro, > and added a specific rules file for openSUSE 11.1 > - updated Debian and RPM binary scripts to point to correct udev > files > 2009/02/07 > - applied Nicolas Vivien's patch to fix the timeout issue on the > Blackberry Storm > 2009/02/04 > - applied Josh Kropf's save module patch > - minor whitespace fixes > - removed unneeded exception (fstream closes files > automatically on exit of scope, or exception) > - fixed compile error with variable sized array... > replaced with std::vector<> instead of memcpy, > in JavaLoader::SaveData() > - commented out an unused variable > 2009/02/03 > - commented out AT+CREG? in barry-rogers.chat, since it causes some > Blackberry devices to reboot > - applied Josh Kropf's eventlog and clear eventlog patches > - capitalized JLEventlogEntry fields, to match other > record classes > - use constant when parsing entry > - renamed a define > - sweeping cleanup of record classes and clarified all constant > conversions i.e. whenever converting between a protocol > constant and an API constant, do it through a function, > in case the constants change on the Blackberry side > 2009/01/31 > - applied patch of binary VSM file from Robert Yaklin > 2009/01/30 > - moved Nicolas Vivien's codfile code into library class API > - this change avoids having packed structs exposed in > installable header files, and gives the library > user an easy way to work with COD files > - added CodFile class > - added JavaLoader::LoadApp(CodFile&) > - cleaned up bjavaloader to use new API > - moved codfile_*_t structs to internal header > - also fixed command line status output when exceptions > are caught (unrelated fix) > - added / updated manpages: bfuse, bjavaloader, btool > - added fuse to build dependencies for rpm and deb packages > - updated web docs, adding libfuse to dependencies page > - added libiconv as well > 2009/01/29 > - applied screenshot and zsh patch from Nicolas Vivien > His git log, as posted to the mailing list, was: > Remove bscreenshot binary > Add screenshot option to bjavaloader binary > Add comments in the code > Use defines beside of integer constant > Use b2hons to be compliant with evrywhat host > ZSH scripts completion > Use BMP structs beside of unreadable buffer ! > His changelog notes: > - add command GetScreenShot to JLPacket class > - add JLScreenInfo to protocol structures > - updated AUTHORS > - minor tweaks to Nicolas's screenshot+zsh patch: > - renamed DIR_ENTRY and SS_ENTRY to DATA_ENTRY constant > as suggested by comments > - fixed some packet size checks > - whitespace fixes > - removed unneeded packet.GetReceive(), since response > buffer is the same as what GetReceive() returns > - used AutoClose for fopen() calls in bjavaloader.cc > for exception safety > - removed manual buffer handling in favour of Data object > in JavaLoader::GetScreenshot() > - removed "unknown packet" exception from Socket::PacketData() > I think it is probably better to handle javaloader-level > errors and codes in the JavaLoader class, not the low- > level Socket class. > - added erase and force erase support to bjavaloader, library, and > zsh > - force erase is experimental > - applied documentation patch from Robert Yaklin on VSM file format > - did slight formatting changes before applying, for 80 > columns, and set file to unix format instead of DOS > - added links to new USB capture logs from Robert Yaklin to > logs.php doc page > - cleaned up src/Makefile.am, regarding non-install headers > - moved Nicolas Vivien's bitmap conversion code into library API > - added bmp.h api, and moved internal structs to > bmp-internal.h > - added size checks and endian conversion to bitmap > conversion code, so this code should run anywhere > 2009/01/27 > - added link Josh Kropf's 'Blackberry development in linux' blog > post to doc/www/index.php > - added USB capture log archive page to www docs > - added config.rpath, needed by new AM_ICONV check > Thanks to Bill Paul for debugging this. > - applied Nicolas Vivien's opensync plugin 0.4x patch > 2009/01/26 > - added template wrapper for iconv() due to casting differences on > various platforms... thanks to Bill Paul for FreeBSD > testing > - added AM_ICONV automake/autoconf check to configure.ac > This should take care of checking for libc or library > iconv installs, and checking the iconv() prototype for > whether > const is needed or not. This removes the need for the > template added earlier. > 2009/01/25 > - applied patch from Josh Kropf making bjavaloader's syntax more > compatible with the windows version, to make cross > platform build scripts possible > - applied Josh Kropf's settime syntax patch > - clarified timezone behaviour in javaloader settime > - added flag to JavaLoader mode class, so StopStream() is called > automatically in the destructor if not called manually > 2009/01/24 > - added another ctor to BadSize exception class > - fixed bug in SocketZero where it checked sequence on packets > from non-zero sockets > - javaloader list mode updates: > - added Dump() code to directory classes > - fixed size issue with JLDirEntry header size define > - fixed size checks in entry parser > - fixed copy before in JLDirectory, now uses vector > - moved bjavaloader's Start/StopStream calls to main() > 2009/01/23 > - removed default argument to CheckSize() so all code specifies > the size it really needs > - fixed bug in ParseData() that didn't copy data to receive arg > - adding additional javaloader commands and refactoring > this is a work in progress: > - added JLPacket class, encapsulating javaloader protocol > - added new defines for more JL commands > - refactored socket class, using PacketData() > for JL, and Packet() for Desktop (still needs > some work to be really clean) > - began implementing new commands (SetTime()) > - thanks to Rick Scott for his reverse engineering effort > in XmBlackBerry, which inspired this code > - refactored packet.cc to make JLPacket even tighter > - updated doc/USB-capture.txt with link to vusb-analyzer > Thanks to Martin Owens for the link. > - added a set of be_*() endian swap functions, for big endian data > JavaLoader code seems to use big endian a lot. > - simplified bjavaloader.cc's command line args to make room for > more > - cleaned up big endian code to use new be_* swap macros > - added protocol and packet code for directory / subdir commands > - added preliminary directory support to JavaLoader class > 2009/01/22 > - reverted experimental handshaking in IpModem, since the > session_key > patch fixed Bill Paul's issue > - finished adding IConverter support to r_task and r_timezone > - removed r_message_base.h from doxygen RecordParserClasses group > - removed unused ClassType variable from Task record class > - refactored Calendar and Task record classes, creating new > RecurBase to handle recurrence data > - removed FIXME in Socket::Packet(), since socket is set in Send() > - cleaned up socket handling and comments in packet.cc > 2009/01/20 > - applied Nicolas Vivien's javaloader patch, fixing constant > defines > and other small fixes > - minor whitespace and spelling fixes to above patch > - changed m_ipmodem.cc to send the session key even without > password > which should help reduce the "special packets" coming from > the device. > Thanks to Rick Scott and Andy Herkey. > 2009/01/17 > - refactored pthread timeout creation into ThreadTimeout() function > - added experimental handshaking to IpModem. This is in response > to Bill Paul's email regarding what he was seeing in the > logs with the special code packets which we dropped > on the floor. This code is to test whether this > handshaking > theory is true. > 2009/01/16 > - made maintainer script make-deb-local.sh more robust > - added list of target binary packages in ReleaseChecklist.txt > - changed to more general Fedora link to Chris Stover's packages > 2009/01/15 > - applied Andy Herkey's ipmodem password patch for the Bold > - minor fixes to above patch, and in addition to above patch: > - use sizeof(seed), instead of type, in case of future > changes > - removed commented out code > - added the null password check back to SendPassword(), > for historical reasons > - removed the null and zero default arguments to > SendPassword(), since they make no sense > - added define for session key size, and added size check > - added missing string.h header to src/m_javaloader.cc > - updated Debian and RPM packages to include bjavaloader > 2009/01/14 > - applied Nicolas Vivien's bjavaloader patch > - small fixups to above patch: > - moved cod.h from installable headers to internal headers > - whitespace edits > - copyright statement tweaks > - added blog URLs to copyright list in cod.h where parts > of the structs appear to come from > - removed unused variables to compile with strict settings > - changed char to unsigned char to fix strict compiler > warnings > - removed duplicated a/b from Socket messages, and moved > read/write text to front, in socket.cc > - added NetDirect copyright notice to bjavaloader.cc, since > it is based on btool.cc > - removed unused boost-specific code from bjavaloader.cc > - more fixes to above patch: (bjavaloader.cc specifically) > - moved cod.h back to installable headers, since tools/ > programs depend on it... this should be fixed > or put in the library as a COD-reading API > so that applications don't need to worry about > packed files, etc. > - fixed potential buffer overflow: removed buffer read used > to skip ahead in file, and replaced with fseek() > - changed SendAppFile() to report errors by exception > instead of by return value, which wasn't checked > anyway... and C++'d the file handle closing > - removed unused StateTableCommand, leftover from btool > - added check for unlikely case of COD file being larger > than size_t, breaking the fread() assumption > for the type == 0xC0DE case > - added checks for fread() failures > - fixed compiler warnings when comparing signed/unsigned, > and changed file operation types from plain int > to size_t/off_t > - changes to src/m_javaloader.cc: > - fixed comments to remove Desktop-oriented notes > - moved variables closer to their use point > - removed the duplicated response Data objects in favour > of just one (each default Data object uses 0x4000 > bytes of heap) > - removed manual endian swap in favour of htobl() > - oops... this size field is actually big endian. > Since all Barry code assumes that Blackberry > protocol traffic is little endian, a special > case needs to be added in m_javaloader.cc for > this field. The special case uses autoconf's > endian define. > - fixed long long problem in endian.h (ISO C++ > compilers don't support long long) and made > the bswap() functions always available > - removed unused CommandType and m_commandTable members > - fixed exception error messages to indicate command # > instead of Desktop message > - changed Packet() call to PacketData() when sending > file size, since there is no command > - added eout() data dump to socket::Packet() unknown send error > - fixed my incorrect off_t/size_t check in bjavaloader.cc > - removed the '0' JL_UNKNOWN4 command from src/protocol.h, > since the PacketData() change above now handles that, > and I don't think that byte is actually a command, but > really part of the uint32_t size > - Also fixed one remaining 0 command, to use PacketData() > 2009/01/09 > - added AT+CLAC note to contrib/modemtest.rb, recording tidbit > from mailing list > - updated copyright dates for 2009 > 2008/12/11 > - fixed missing slash in config sample in doc/www/sync.php > Thanks to sourceforge tracker #2413683 for reporting > 2008/12/01 > - added more notes to doc/USB-capture.txt, documenting my kernel > setup for capturing USB traffic, and added size-limiting > kernel patch > - added TODO list item for the Blackberry Storm > 2008/11/30 > - moved Product ID constants into a single place (src/common.h) > and updated bcharge > - added Product ID 8007 for the Blackberry Storm... needs testing > 2008/11/27 > - applied ppp chat script patch for AT&T Cingular from > Richard Esplin > - added new chat scripts to debian and rpm packages > - added known sources of additional Barry binary packages to > index.php > - turned on call/caller graph support in Doxygen output > 2008/11/23 > - applied 'patch' of Martin Owens' Content Store documentation > - updated AUTHORS > 2008/11/22 > - large refactoring of all message/email record classes, and more > There is now a MessageBase base class, which combines > common email record parsing code in one class. This base > class is reused by all the specific messages record > classes. > This commit also includes support for multiple email > addresses per email, as well as giving IConverter support > to all message classes. > - deleted empty *.cc files from project > - updated TODO file, removing opensync i18n bug > - added TODO item for password support in Opensync... try to avoid > storing the password in the sync config in plain text > - fixed dll export bug on ErrnoError exception class > - added BadPackedFormat exception for internal use > - ServiceBook: added IConverter support, and general cleanup > Refactored ServiceBook class and added some experimental > template functions to generalize the FieldLink table > processing. If possible, this needs to be expanded > to the general record classes, to reduce the amount of > code duplication during processing of FieldLinks and > CommonField structures. > Also, ServiceBook deals with packed fields, which should > also be extrapolated for general use, and reused if > possible for the new "Calendar - All" Blackberry databases > that are coming out on the new devices and firmare. > 2008/11/21 > - updated man/btool.1 with new -i option > - fixed bug where opensync plugin didn't pass IConverter to Desktop > - added IConverter support to r_calendar.cc > - renamed utf8Needed to iconvNeeded, for consistency > - added IConverter support to r_folder.cc > - added IConverter support to r_memo.cc > - added IConverter support to r_message.cc > - added EmailAddressList, in preparation for multi-address email > 2008/11/20 > - changed Parser base class to pure abstract... this is so that > future API changes to the Parse base class will be > detected by the compiler as errors > - added NullParser class, which performs the old Parser function > - changed all FieldLink arrays to static > - added new IConverter class to support converting the Blackberry's > WINDOWS-1252 charset to/from UTF-8 and more > - fixed bug in IConverter that saved end of string instead of > beginning > - reorganized barry.h app header and record.h header, so that > internal builds have fewer dependencies > - large parser/builder API change, adding IConverter support > This change starts at the top (Mode::Desktop) which > takes an optional IConverter object for parsing and > building conversions. The Desktop class passes > this IConverter object to Packet, Parser, and Builder (and > thereby, the corresponding Record classes) as needed. > This adds an extra pointer to a fair number of calls > in the stack, but maintains flexibility without putting > too much of a burden on Parser writers, which could be > application writers. > The primary use of this architecture is to pass in an > IConverter configured to convert to UTF-8... this is > basically the only useful conversion method, but others > are possible if you want to play. > - first record class to support string conversions: Contact > - reverted the barry.h and record.h reorganization, since > the gui depends on it > - reorganized barry.h and record.h again, making inclusion of > the record headers dependent on whether it is a library > build or not > - updated the gui with the parser/builder API change > - added IConverter to opensync plugin, set to UTF-8 > 2008/11/13 > - added comment to src/r_contact.h deprecating the Phone field > - updated Doxyfile to 1.5.1 > - added Christopher D. Stover to AUTHORS file as Fedora contact > 2008/11/03 > - forgot #include <string.h> for memcpy/memset in tools/bfuse.cc > - moved bfuse install point from /usr/sbin to /usr/bin > 2008/11/01 > - added FUSE filesystem "driver" > 2008/10/31 > - commented out the info.subsystem match level in HAL FDI file... > this should let the FDI work on old and new systems. > HAL version 0.5.8 (used by Debian Etch) has the deprecated > info.bus property, and doesn't have info.subsystem yet. > - added HAL FDI scripts to rpm/barry.spec > - fixed tools/Makefile.plain > 2008/10/30 > - added support for multiple work/home phone numbers to the > opensync plugin. > - Thanks to Richard Esplin for his test data. > - This fix has a conflict when a cell phone number is > marked as the preferred contact as well. > See > http://sourceforge.net/mailarchive/message.php?msg_name=20081030231045.GA28989%40foursquare.net > for the full details. > - This fix also changes the VCARD output... pager phone > numbers are given the TYPE code of "pager" > instead of "msg". > - Added OtherPhone field as well > 2008/10/22 > - applied HAL FDI scripts and configuration from Martin Owens > - updated AUTHORS file > - added HAL FDI and python script to Debian package install > 2008/10/18 > - updated copyright notice and help text in contrib/modemtest.rb > - changed barry-verizon.chat script according to Andy Herkey's > recommendation... this worked much better for Michael L. > Stokes, since his provider was getting stuck on the ATZ > 2008/10/16 > - applied Andy Herkey's modemtest.rb ruby script to contrib/ > from his Aug 9, 2008 email to the mailing list > - applied Andy Herkey's new modemtest.rb script to contrib/ > from his Oct 8, 2008 email to the mailing list > - added password pass-through code for pppob -s mode > - clarified front page www doc, adding date of last release > and license > - cherry-picked changes from Jose Carlos Garcia Sogo's debian git > repo > - debian/rules was stripping wrong library names > - FSF address was incorrect (also updated other source > files throughout the tree myself) > - more complete debian/copyright file > 2008/09/24 > - version bump: > - configure.ac > - src/Makefile.am > - src/version.cc > - gui/src/BackupWindow.cc > - Doxyfile > - rpm/barry.spec > - debian/changelog > - opensync-plugin/src/barry_sync.cc > - added link to doxygen documentation to www index.php > > > ------------------------------------------------------------------------------ > Let Crystal Reports handle the reporting - Free Crystal Reports 2008 > 30-Day > trial. Simplify your report design, integration and deployment - and focus > on > what you do best, core application coding. Discover what's new with > Crystal Reports now. http://p.sf.net/sfu/bobj-july > _______________________________________________ > Barry-devel mailing list > Barry-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/barry-devel > ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Barry-devel mailing list Barry-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/barry-devel
