The loop for randomizing a list in the dissolve transition is poor.
It randomly chooses two element and swaps them, and does this N/2 times.
Thus a total of N elements are moved.
It is probable that several elements will be swapped multiple times,
leaving it probable that several other elements will not be swapped at all.
This fact is quite visible when transitioning between two very different
slides (e.g. 2 photos). While there is a lot of randomness, there is a
clearly perceptible top-to-bottom 'wipe' happening at the same time.
So change the loop to provide true randomness.
We randomly choose an element for the last position.
Then randomly choose one of the remaining elements for the second
last position, etc.
Signed-off-by: NeilBrown <neilb@suse.de>
* resolved two conflicts caused by lineend changes in
canvashelper_texturefill.cxx and implrenderer.cxx. Took the cws
version, which uses new texture fill semantics, and fixes clipping
bug
* binned file header changes from head for deleted files
2009-10-13 15:51:48 +0200 fs r276867 : allow building with 'nodep'==''
2009-10-11 22:39:56 +0200 fs r276820 : silence a compiler warning in a file only used when VERBOSE!=""
2009-10-09 14:37:43 +0200 fs r276807 : let not override the non-presence of $VERBOSE the given command line arguments
2009-10-09 14:36:25 +0200 fs r276806 : also deliver spirit/home/classic/debug/impl
2009-10-08 13:38:07 +0200 fs r276789 : not that many line feeds in --show mode
2009-10-08 13:13:01 +0200 fs r276788 : minor adjustments requested by hjs:
- replaced space/tab mixes at beginning of lines with mere tab
- made "nothing to update" message for the ZIPALLTARGET target more prominent
- removed useless "echo > /dev/nul" statements
2009-10-07 13:32:12 +0200 fs r276753 : #i105585#
2009-10-07 11:31:59 +0200 fs r276742 : #i10000# missing dependency between stoc/security and stoc/util
2009-10-06 22:59:59 +0200 fs r276729 : CWS-TOOLING: rebase CWS buildverbosity to trunk@276699 (milestone: DEV300:m61)
2009-09-29 12:04:58 +0200 fs r276518 : #i10000#
2009-09-29 12:04:25 +0200 fs r276517 : oops
2009-09-29 12:01:07 +0200 fs r276516 : #i84497# removed some more (non-diagnostic) 'echo ------' directives
2009-09-29 11:59:29 +0200 fs r276515 : #i84497# removed some more (non-diagnostic) 'echo ------' directives
2009-09-29 11:52:32 +0200 fs r276514 : #i84497# removed the various 'echo ------------' directives from verbose mode, as per hjs' request
2009-09-29 11:44:16 +0200 fs r276513 : do not duplicate IDLC call commands with different verbosity switches
2009-09-28 21:43:50 +0200 thb r276502 : #i84497# fixed potential recursive macro def + extra ls
* solenv/inc/settings.mk: now setting via VERBOSE!:=, removes warning
and my spurious "recursive macro definition" error
* solenv/inc/tg_app.mk: one extra ls silenced for app target (and
quiet mode, that goes without saying)
2009-09-23 08:57:01 +0200 fs r276366 : use ULFEX_VERBOSITY instead of duplicating the ULFEX call
2009-09-21 11:42:26 +0200 fs r276320 : ignore output paths
2009-09-21 11:04:27 +0200 fs r276318 : silence another compiler warning, which only hits us when actually using this (debug) file, which is the case only when slideshow is compiled with a env variable VERBOSE, thus came up in CWS buildverbosity
2009-09-21 10:30:14 +0200 fs r276313 : make some output depend on VERBOSE==TRUE, not COMMAND_ECHO=""
While both are equivalent at the moment, COMMAND_ECHO finally is an implementation
default of the VERBOSE flag only, so better rely on VERBOSE as the primary verbosity flag
2009-09-21 09:56:57 +0200 fs r276311 : forgot to re-generate from tg_zip before committing
2009-09-21 09:52:11 +0200 fs r276310 : #i84497# even less verbosity
2009-09-17 11:02:47 +0200 fs r276232 : #i84497# verbose implies VERBOSE nowadays, and VERBOSE==FALSE should not lead to -DVERBOSE
2009-09-15 22:59:37 +0200 fs r276189 : #i105022# copy fix for this P1 into this CWS
2009-09-15 11:56:35 +0200 fs r276165 : CWS-TOOLING: rebase CWS buildverbosity to trunk@276043 (milestone: DEV300:m58)
2009-09-14 17:45:10 +0200 fs r276137 : #i84497# don't duplicate zip lines w/ and w/o -q switch, use a variable instead (maintenance)
2009-09-14 15:01:33 +0200 fs r276124 : #i10000#
2009-09-11 23:58:46 +0200 thb r276083 : #i84497#: More quiet-build fine tuning
- silenced rsc for real (properly filtering options for
cpp, and a pretty brutal amputation of the tool blurp,
which would have needed cmd opt parser duplication)
- silenced deliver.pl
- silenced checkdll.sh
- silenced zip via -q (in quiet mode)
- silenced various idl, resource, transex whatever tool,
passing appropriate options down to them '-QQ' sometimes
- silenced dmake, pointless blurb that something does *not*
need update removed
- silenced the old starview idl compiler, to not output
tool's name & progress chars in quiet mode
2009-09-11 21:13:59 +0200 thb r276081 : #i84497#: More bits on the quiet mode of the build.
The bulk of the changes is disabling those annoying
"echo ------------------" lines for the quiet build,
which has the stretch goal of outputting *exactly*
one line per file compiled/linked/processed.
Apart from that, silenced a few especialy annoying
module-specific makefiles (basically adding
$(COMMAND_ECHO) in front of a gazillion rules).
Additionally, slightly tweaked what idlc regards
verbose, normal, and quiet mode; this was to have it
echo exactly one line per idl file processed (the
fact that quiet mode did not echo *anything* for idl
files was a bit too much for my taste)
2009-09-04 09:14:35 +0200 fs r275777 : don't emit the link parameters if VERBOSE!=TRUE
2009-09-02 10:31:15 +0200 fs r275700 : #i10000#
2009-09-02 08:44:14 +0200 fs r275697 : update svn:ignore to ignore the output paths
2009-09-02 08:40:54 +0200 fs r275695 : #i84497# less verbose output during build, unless a dedicated '-verbose' switch is given
2009-09-02 08:40:28 +0200 fs r275694 : GRAPHITE is missing in the BUILD_TYPE
2009-09-02 08:40:05 +0200 fs r275693 : #i84497# less verbose output during build, unless a dedicated '-verbose' switch is given
2009-09-02 08:39:25 +0200 fs r275692 : #i84497# less verbose output during build, unless a dedicated '-verbose' switch is given
2009-09-02 08:39:02 +0200 fs r275691 : #i84497# less verbose output during build, unless a dedicated '-verbose' switch is given
2009-09-02 08:38:09 +0200 fs r275690 : #i84497# less verbose output during build
2009-09-02 08:37:06 +0200 fs r275689 : #i84497# don't emit that much noise, unless a '-verbose' switch is given
2009-09-09 10:30:24 +0200 fs r275970 : #i100127# provided by rene: check whether system-boost's function.hpp compiles without exception support
2009-09-04 21:58:00 +0200 fs r275832 : indention
2009-09-04 21:57:10 +0200 fs r275831 : #i10000# type in boost/spirit check. Thanks to thb
2009-09-04 10:58:54 +0200 fs r275787 : #i100127# provided by tono@openoffice.org: mingw needs using directives for std::min/max
2009-09-04 10:31:54 +0200 fs r275782 : #i100127# (approved by hjs) on Solaris, set GNUTAR to /usr/sfw/bin/gtar, this is available on baseline machines (Solaris 10.x), but not necessarily available in LD_LIBRARY_PATH
2009-08-19 21:47:55 +0200 rene r275161 : #i10000# check for the header actually used now
2009-08-17 21:55:31 +0200 fs r275073 : CWS-TOOLING: rebase CWS boost134 to trunk@275001 (milestone: DEV300:m55)
2009-07-16 09:39:25 +0200 fs r274031 : #i100127# extract download/% with GNUTAR instead of TAR
2009-07-14 09:18:15 +0200 fs r273960 : CWS-TOOLING: rebase CWS boost134 to trunk@273858 (milestone: DEV300:m52)
2009-07-13 09:55:14 +0200 fs r273919 : #i100127# use boost 1.39, not 1.34
2009-07-13 09:45:57 +0200 fs r273918 : use GNUTAR when extracting download/%.tar.gz, not TAR (necessary with boost 1.39 tarball)
2009-07-10 21:10:52 +0200 fs r273905 : #i100127# spirit now part of boost
2009-07-06 12:12:13 +0200 fs r273732 : #i10000# lngconvex, used in src/wizards, is built in shell => added missing module dependency
2009-07-02 15:34:51 +0200 fs r273660 : #i100127#
2009-07-02 13:35:24 +0200 fs r273645 : #i100127# MSVC needs some hinting when using boost::bind on (some) member functions
2009-07-02 10:33:59 +0200 fs r273625 : removed unused include (lead to an internal compiler error with MSVC and boost 1.34+)
2009-07-02 09:44:35 +0200 fs r273619 : change some include orders, prevents 'internal compiler errors' with MSVC
2009-07-01 09:19:25 +0200 fs r273547 : #i100127# who the heck is defining max as something unusable?
2009-07-01 09:17:28 +0200 fs r273546 : #i100127# defining min/max is causing trouble now
2009-07-01 09:15:40 +0200 fs r273545 : #i100127# reverted previous patch, and properly fixed the problem by defining BOOST_MEM_FN_ENABLE_CDECL, and disabling warning 4180
2009-07-01 09:14:23 +0200 fs r273544 : #i100127# re-disable warning 4180, still needed for some files
2009-07-01 09:07:33 +0200 fs r273543 : #i100127# reverted previous patch, and properly fixed the problem by defining BOOST_MEM_FN_ENABLE_CDECL
2009-06-24 11:00:32 +0200 fs r273319 : oops, re-introduced some inc locations
2009-06-23 22:10:02 +0200 fs r273304 : #i100127# disable warning C4180
2009-06-23 22:08:36 +0200 fs r273303 : #i100127# disable warning C4180
2009-06-16 12:26:49 +0200 fs r273015 : #i10000# (provided by mst@openoffice.org)
2009-06-15 09:10:27 +0200 fs r272962 : #i10000# use -r instead of -a when copying the include files, this prevents the spurious 'failed to preserve ownership' errors
2009-06-11 23:55:06 +0200 fs r272902 : #i100127#
2009-06-11 23:54:47 +0200 fs r272901 : #i100127#
2009-06-11 23:54:05 +0200 fs r272900 : make compile on unxsol[i|s]4
2009-06-10 10:12:22 +0200 fs r272798 : #i100127# don't deliver from within unxlngi6/misc/build/boost, but from a 'normalized' location unxlngi6/inc
2009-06-10 09:48:00 +0200 fs r272795 : #i100127# use 1.34.1 unconditionally on all platforms
2009-09-01 #i104566# also use syslibroot/isysroot for external components
2009-08-28 #i104568# clean up duplicated definitions of OBJCXXFLAGS
2009-08-28 #i104567# add ccache-detection to configure (to skip to objcxx-flags)
2009-08-28 #i104566# use Mac OS X 10.4-SDK to ensure backwards compatibility
use MACOSX_DEPLOYMENT_TARGET and sysroot/syslibroot switches
ignore "-isysroot /Developer/..." arguments for OOo's cpp
2009-04-14 13:09:13 +0200 sb r270758 : CWS-TOOLING: rebase CWS sb107 to trunk@270723 (milestone: DEV300:m46)
2009-04-06 14:11:54 +0200 sb r270545 : #i100884# improve pagein effectiveness: wrap system libs in --as-needed on unxlngi6/unxlngx6 to avoid needing them where unneeded; link pagein as C program to avoid unneeded C++ runtime dependency; reorder pagein and javaldx in soffice startup script so that javaldx benefits from pagein
2009-03-31 16:40:23 +0200 sb r270287 : decrease collisions in hashCode computation
2009-03-30 11:48:39 +0200 sb r270207 : #i100668# corrected explicit template instantiation
2009-03-26 14:16:26 +0100 sb r270078 : #i100408# inadvertently dropped libexslt dynamic library in previous rev 269789
2009-03-26 10:52:51 +0100 sb r270058 : #i100576# fixed erroneous modification of RPM in previous rev 270057
2009-03-26 10:46:36 +0100 sb r270057 : #i100576# Enable HAVE_LD_HASH_STYLE in sdev300.ini for unxlgni6 and unxlngx6; corrected unxlngx6.mk so that HAVE_LD_HASH_STYLE actually has an effect on ultimate LINKFLAGS value.
2009-03-25 16:53:39 +0100 sb r270038 : #i85679# applied speed-symbolic-functions.diff provided by pmladek; re-ran autoconf; enabled HAVE_LD_BSYMBOLIC_FUNCTIONS for unxlngi6 and unxlngx6 in sdev300.ini
2009-03-25 15:29:12 +0100 sb r270031 : #i100408# missing fixes in addition to -c 269789 (to actually only put single instances of certain libraries into installation sets)
2009-03-25 15:17:10 +0100 sb r270029 : #i10084# revert masterfix additions of libraries to APPnSTDLIBs (during -r269000:269199) and instead use -rpath-link to avoid linker warnings on unxlngi6 and unxlngx6
2009-03-25 14:15:23 +0100 sb r270023 : added svn:ignore
2009-03-25 10:25:08 +0100 sb r270002 : #i10000# copied over trunk -c 269820
2009-03-23 10:16:00 +0100 sb r269858 : CWS-TOOLING: rebase CWS sb107 to trunk@269781 (milestone: DEV300:m44)
2009-03-20 11:22:27 +0100 sb r269789 : #i100408# reduce (multiple copies of) libraries delivered to solver; for that to work fine, deliver symlink feature got improved to only create a symlink if the original file exists
2009-03-20 09:21:32 +0100 sb r269782 : #i100396# replace s(HTML|RTF)_xxx declarations with OOO_STRING_SVTOOLS_(HTML|RTF)_xxx macros to reduce symbolic relocations at load time (at least on ELF systems).
2009-03-19 11:34:56 +0100 sb r269734 : #i100348# added VISIBILITY_HIDDEN=TRUE so that symbols from the resulting archive are not erroneously exported from dynamic libraries including the archive
2009-03-19 11:32:20 +0100 sb r269733 : #i100348# as a prerequisite for following changes, brought jpeg-6b.patch into "dmake create_patch" format and removed application/octet-stream svn:mime-type
2009-03-12 10:39:08 +0100 sb r269370 : duplicated cws/odfmetadata2/solenv/inc/target.mk -c 268831
2009-03-04 17:25:16 +0100 sb r268849 : #i99880# missing AUGMENT_LIBRARY_PATH
2009-02-25 17:54:34 +0100 sb r268459 : CWS-TOOLING: rebase CWS sb107 to trunk@268395 (milestone: DEV300:m42)
2009-02-25 09:23:15 +0100 sb r268413 : #i99584# avoid undefined operations on nOffset (patch by cmc)
2009-02-23 10:02:25 +0100 sb r268344 : #i99519 removed unnecessary library dependencies
2009-02-18 17:54:24 +0100 sb r268250 : avoid warnings about format specifier and argument mismatch (on 64bit debug builds)
2009-02-18 17:50:34 +0100 sb r268249 : avoid erroneous warning that control reaches end of non-void function
2009-02-18 17:33:53 +0100 sb r268248 : sdext/source/pdfimport/misc/pdfihelper.cxx: #include <vcl/canvastools.hxx>
