It turns out that the ordinal number adjustment that was added for
fdo#34548 is no longer necessary.

Change-Id: I2c491a008500b6a269b6f05a2a72f5aa70edecb2

Change-Id: I4290659c506ab11825b4c2f36952214b6254252c

Change-Id: I240c8c940d7d3e1310c4ee33911e8c7019e67060

Change-Id: I08f17dd9cc092206083ff41bbbc178e0322e86d0

Change-Id: Ia8f2c9f1c1c284708a2cbde379197ec6ba58742f

Change-Id: I2a4ddb2e3a3ff629243dba46880b8c986dfde2dc

Change-Id: I1e3b1d4b7d596630ea9b804ded911a2e48b970c9

Change-Id: I4ccf1cc5609ed4f2c70dbf396264e6caf16573c3

The problem is that SwUndoDelete will move the fully selected nodes to
the UndoNodes but it leaves bookmarks with their SwIndex pointing to the
deleted nodes.  The SwNodeIndex are corrected by SwNodes::_MoveNodes()
so they point to a different node than the SwIndex.

This only happens if only one position of the bookmark is inside the
deletion range; if both are, the bookmark will be deleted by
SwUndoSaveCntnt::DelCntntIndex().

Also joining the 2 start/end nodes of the selection will accidentally
correct the bookmarks but only if it happens to delete the end node.
(and apparently there is also a DeleteRange method that doesn't join)

Change-Id: I91ec362bb833328f8d681fd9458cb915c4efb267

... with a more useful assertion.

Change-Id: I43f0a3e123d9dc9f9fd9548e84b8619c7d4f1262

Change-Id: I3a012cfd9ba8c8c331d025f56f590a74941f71dd

Change-Id: I5f2393b13729ab43ad2cfd4a3f960a507e3e608b

Change-Id: I5dc00f7b03b51916b1331c4ba64064a5aa5930c9

Change-Id: I129f3dc1597035664e4ff284276cb0d49a560ab5

If these are called on a value that is of a different type than what is
checked in the children() method, nothing is printed except exceptions
about non-existent _iterator.

GDB can figure out the dynamic type itself and call the pretty printer
only if it's actually a UnoMark so don't register it for IMark.

Same thing for SwUnoCrsr, there is already a class for it.

Change-Id: I00238f1e3f01741d607eec9d47671302b858b965

Change-Id: Ieb1e104e8147376405e90ecada308fee05321dcd

... before calling into BASIC.

Change-Id: I51a59fe5ca9b15ff7ec509c052964523d87a9790

Change-Id: Ic1fbb36a509ec1dce63f1c7fa4a8955a4780517d

Change-Id: Icb0109b954b9a7710309cc9bca30309a6ec90a73

Change-Id: Ibaaa9ce5e5469b960d2b3dee1d40c536e5ae0739

Project: translations  add7a7e0967a3fc04666ba061d81d1ba9b326554

Change-Id: Icbab4d030877f978babfc51f984fb4793b60f681

The exception thrown by at() is not treated meaningfully anyway.

Change-Id: I8c243541271d7da76b2a9b102e058b5f2d1ad6d3

and some contain a column in the second child

Change-Id: Ifd69758336233ed0233120b3315d4f33655fa994

Change-Id: I6678502472f646ae770be4d17d85f2a02fba8f4b

Change-Id: I4f155f8a28afa6a16b1859b254a339d8374a769b

Turn off AutoCalc while constructing it and then calculate it when ready.

Use a smaller number for the minimum formula group size required for OpenCL to
be used while calculating it. Use less hardcoded magic numbers in the code.

Change-Id: I6f2c3b1035947a542776d6087bd229cb9c2892fb

Change-Id: Ia4ac90c369a8e33ddefc6a87f877b98491efcede

Change-Id: I25d681b4b1e35af49b6631b2c3de2ccc296b9b42

Change-Id: Ifb0a277a27613ea1a771a8bb3b8c784b9eca4f2f

Change-Id: I25b4ccdc2f50ba89687e1f4c871f286251a28e21

...by fully decoding them via DECODE_WITH_CHARSET (where charset defaults to
UTF-8) instead of merely normalizing them to IRIs via DECODE_TO_IRI.  For
example, entering

  http://example.com/foo%2Fbar

into the Hyperlink dialog would have resulted in a corrupted URI

  http://example.com/foo/bar

Change-Id: Ic37859afa59c3fdc3280cf4803d4334026e1f336

...in some places where it is obvious that it does not hurt that for an empty
vector the obtained pointer is not necessarily a nullptr.

Change-Id: Id5d66b1559ca8b8955d379bcdbfae6986ef46a51

Change-Id: I57c49172fa5bb19968bf217285d0cd9222cc3530

Change-Id: Id502a36ece4ecc4551ee67f1818fb9886d6e0ce5

Change-Id: I9c3d5b94d22566d63f49d9dcd73a97df97e9802e

Needed for the NORMDIST() OpenCL implementation test doc.

Change-Id: I40f0d5d7ccf5963fa47d6648873ab23f8adf7aea

i.e. load doc keeping setting text of "LINE" + ok as each popup appears
will crash eventually

Change-Id: Ibc4bb8fe65bd5dd380c0fe833e5ece2b39de5dbe

Just temporary (haha, let's see...), the Right Thing would be to turn the
mnOpenCLMinimumFormulaGroupSize down to some small value for the duration of
the test document's construction and calculation, and then return it to its
prevous value once the document has been dismissed.

Change-Id: I52e9813af1ceb2410ef74723fadee21b89cffe47

Change-Id: Idd27688055a08ffe9a6970c28cd7da6f6e1a7f66