|
Apache OpenOffice (AOO) Bugzilla – Full Text Issue Listing |
| Summary: | Optimizing algorithm of creating pivot table cache when loading a xls file with pivot table | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Calc | Reporter: | Tan Li <litan.test> | ||||||
| Component: | open-import | Assignee: | AOO issues mailing list <issues> | ||||||
| Status: | CLOSED FIXED | QA Contact: | |||||||
| Severity: | Normal | ||||||||
| Priority: | P3 | CC: | arielch, leiw, liushenf, pfg | ||||||
| Version: | 3.4.0 | ||||||||
| Target Milestone: | 4.0.0 | ||||||||
| Hardware: | All | ||||||||
| OS: | All | ||||||||
| Issue Type: | ENHANCEMENT | Latest Confirmation in: | --- | ||||||
| Developer Difficulty: | --- | ||||||||
| Attachments: |
|
||||||||
|
Description
Tan Li
2012-09-07 07:56:33 UTC
Created attachment 79350 [details]
patch for this issue
Created attachment 79353 [details]
update previous patch
"leiw" committed SVN revision 1381946 into trunk: #i120847 Optimizing algorithm of creating pivot table cache when loading a xl... Confirm it Submit in revision 1381946 Reopened. Changes break the build on Linux: Checking DLL ../unxlngi6.pro/lib/check_libscd.so ...-rwxrwxr-x 1 buildslave20 buildslave20 62785 Sep 8 02:03 ../unxlngi6.pro/lib/libscd.so ../unxlngi6.pro/slo/column3.o: In function `ScColumn::FillDPCache(ScDPTableDataCache*, long, long, long)': column3.cxx:(.text+0x3f08): undefined reference to `unsigned char ScDPTableDataCache::AddData<false>(long, ScDPItemData*)' collect2: ld returned 1 exit status dmake: Error code 1, while making '../unxlngi6.pro/lib/libsc.so' See bug 120854 "leiw" committed SVN revision 1383822 into trunk: #i120847 Resubmit to take back some codes which is roll back to fix a build b... Resubmit to take back some codes which is roll back to fix a build break bug 120854 (In reply to comment #0) > Optimizing algorithm of creating cache: Current algorithm will visit all > cells one by one to generate cache, need binary search in each column. New > algorithm will create cache for each column Previous algorithm use "new ScDPItemData( pDoc, nRow, nCol, nDocTab )" for each cell to construct cache item "ScDPItemData", this will call "ScColumn::GetCell( SCROW nRow )" ,"pDoc->HasValueData( nCol, nRow, nDocTab )", "pDoc->HasData( nCol,nRow, nDocTab )" , and all these three function use binary search to locate the cell and its properties. Too many redundant search is low efficient. In our solution, for one column, only using binary search for one time to locate cell in start row, for each "ColEntry pItems[nIndex]" in column，using "ScDPItemData( sal_uLong nNF, const String & rS, double fV, sal_uInt8 bF )", or "ScDPItemData( const String& rS, double fV, sal_Bool bHV, const sal_uLong nNumFormat, sal_Bool bData )" for construction; and for each cell between "ColEntry pItems[nIndex]"，using"ScDPItemData()" for construction, and all of these function avoid binary search. "pCell->HasValueData()", "pCell->IsBlank()" are used for replacement. ScDPTableDataCache::AddData(...) is also modified to avoid redundant number format process. Hello; This change uses headers that are not available in modern versions of boost (1.48). Boost is being updated because the newer versions have many bugfixes and work with clang. Please try to use more portable headers or revert until after we finish updating boost. False alarm: I just build the latest version using the external boost 1.48 and the build was successful. The problems building with an updated internal boost are unrelated. (sorry) set Target Milestone to AOO 3.5.0 for PM purpose. |