If it's named nLast, then it's supposed to point to the last character,
which, with an empty string, logically means it should point before
the start of the string.

Change-Id: I248fbe80c1c4b1c2c8d39f0bcf772f9b85415e4a

Change-Id: I2bed0272eade44ab988f2cd9becb1f8ef0f232a9

Change-Id: If623e9a0d909ce6a7e260b0d4beed3fb0c74cf8b

Change-Id: I1d922e5b3c7959022b6f57c33b9a4976a16305f6

Change-Id: I627c2641dcebea1daa5a9264d8b4a5771bf8155f

Change-Id: Idd467f2963099e2ae6bb76f405124dfa32c8cc9a

Change-Id: I3101b9f582861bc88ab694c551e66079f133e3e0

Change-Id: Id495b96f6f496f4ae16d9c9a3697f94c7b3f305c

Change-Id: Ie1dd9091e4d8ee09c9a75eecf28fd6cd06ea1839

Change-Id: I041feabb7ee94b33caebb67a380b8f3694d09e82

Change-Id: I4466b57514352620fd26072544ec6e50bf08708c

Some user do not understand that Exit is for exiting all libreoffice modules and not exiting the current window

Change-Id: If3a88a1e7f79c5c19f3011ba65dde3cda11462a4

Change-Id: I0249d42f06f9c1168c2c0c59af927c4a0535c39e

Change-Id: Ib9cebcf3680e1761c25be03715423845de78bb92

like 3d3b3f65

Change-Id: I6e80717de009e8a3a89ffc80cb945cc832917f8c

Change-Id: I412cd836ab248ed7c247e1125ff52dc3263604d2

Change-Id: I089642eaa1465629575a55d623eea79427bc2871

STR_COLUM_DT_AUTO is unused too.

Change-Id: Icd14e697842ec7c78131a76daa683aae1b5dfc3f

Change-Id: I4c3550bcc214bd675c9f86b95597449255f6154a

...as some client code catches just the former and thus fails now.  (This was a
regression introduced with the recent cppuhelper/source/defaultbootstrap.cxx.)

Change-Id: I8306797f8331d894ab4e7695478e3824e9f79197

Change-Id: Iab813f5288af1f0e054c022c4e4a99b92c7ce1ce

Change-Id: Iad1f2323c91c91a6f8e43414cddb78cf7e739a0d

Change-Id: I18655247f0e836cefd2791e1c78cf9f49e396d02

Change-Id: I00e5e4f959c351004d8c1181c124cc6a1058b666

Change-Id: I34f229284611285e10a192f80b14b1e8b3fa618c

See http://svn.boost.org/svn/boost/trunk r76133 "Fix threading detection in
GCC-4.7 experimental."  Without it, BOOST_DISABLE_THREADS would be defined when
compiling with GCC 4.7, which in turn would cause
boost/smart_ptr/detail/sp_counted_base.hpp to include a single-thread--mode
boost/smart_ptr/detaul/sp_counted_base_nt.hpp which is not suitable in a
multi-threaded scenario.

Change-Id: Ia30a5c2c241e2e135858a5cab3dd855d05db50c8

Change-Id: I4c99129cabe70f17aa223cf8ec0ae1529188b6b7

Change-Id: I37cee7027f674003ad4a600a4a36fc1a9eec2dc3

the original apache commit is quite ugly and should not be used!!

Change-Id: Id02d2cf474e49b97bfe285fe76867d1b1d944ec3

Change-Id: I43b6558de6e92b0fe40a38b5f2417bebad2b7f8b

-This also re-enables ScFiltersTest::testFormatsODS, but disables checks
 for row height import which was causing win tinderbox failures.
-If checks for column width cause win tinderbox failures,
 we may need to temporarily disable them too until we find out
 cause of failures.

Change-Id: Ibfef98d412cf0a25b98d739ad53308dac09ed370

Change-Id: I7e6f495eb56c0d6004c2e427b7fe9c3c09c7206f

Change-Id: Iccbf66d8838fcfa2f2c620dd768aba75f9e071f4

MSVC 2008 refuses to compile this unless it gets a const_iterator, and i
have no idea why.

Change-Id: I08a6916313ae256388309742604dfb5940b42b9f

Change-Id: I01756622dd7700d3918d156f118cd69c8a15879a

In particular com/sun/star/wizards/ui/FilterComponent calls getSelectClause before calling getFromClause, and then all hell breaks loose: composedCommandNames is empty, thus cannot find the proper Alias column name, thus the column names in the select list were not properly escaped, ...
We have just made getFromClause quadratic instead of linear, but:
1) I do not think this would be a problem (small datastructures)
2) If it is, rather use a hashmap or something like that, wich will also make getSelectClause faster

Also make the fallback case of "unknown table" more robust: escape the table name (if any) and column name!

Change-Id: I474adc51fc6378d836bd5865d9eb9505983dcbc5

Change-Id: I404072caf6ddab0ed833586066507c7d332fcea4

Change-Id: Ia465bc3b4bb4c075d235ccd14cac0248b198d470

Change-Id: If6cbb99b59b14fc7911c0fe76f44ac21b0db5042