Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | When processing an "ORDER BY ... LIMIT" that does not use an index, check whether or not a record may appear in the final result set before adding it to the temp b-tree used for sorting. |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | trunk |
Files: | files | file ages | folders |
SHA3-256: |
0fcfc36ceb820fc70136b799a0405fe9 |
User & Date: | dan 2018-04-26 17:43:35.709 |
Context
2018-04-26
| ||
18:34 | The previous fix for ticket [d85fffd6ffe856092ed8da] in check-in [0a514e62ad1ebe5c12da8dae] did not completely address the probably in that it only worked for cases where the OP_SCopy that loaded the register was the last instruction in the sequence for the expression, which is not necessarily the case for expressions like CASE...END. This revision prevents the registered that will be recomputed from being cached in the first place. (check-in: 9fd0faf517 user: drh tags: trunk) | |
17:54 | Merge latest changes from trunk. Including the "ORDER BY ... LIMIT" optimization. (check-in: d8ae7ba083 user: dan tags: begin-concurrent) | |
17:43 | When processing an "ORDER BY ... LIMIT" that does not use an index, check whether or not a record may appear in the final result set before adding it to the temp b-tree used for sorting. (check-in: 0fcfc36ceb user: dan tags: trunk) | |
16:13 | When processing an "ORDER BY ... LIMIT" that does not use an index, check whether or not a record may appear in the final result set before adding it to the sorter. (Closed-Leaf check-in: 71bf91c218 user: dan tags: sorter-limit-opt) | |
15:50 | Ensure that new.* values of an UPDATE do not get clobbered after the BEFORE triggers run when unmodified columns of the row being updated are reloaded. Fix for ticket [d85fffd6ffe856092ed8da] (check-in: 0a514e62ad user: drh tags: trunk) | |
Changes
Changes to src/select.c.
︙ | ︙ | |||
575 576 577 578 579 580 581 | SQLITE_ECEL_DUP | (regOrigData? SQLITE_ECEL_REF : 0)); if( bSeq ){ sqlite3VdbeAddOp2(v, OP_Sequence, pSort->iECursor, regBase+nExpr); } if( nPrefixReg==0 && nData>0 ){ sqlite3ExprCodeMove(pParse, regData, regBase+nExpr+bSeq, nData); } | < > | 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 | SQLITE_ECEL_DUP | (regOrigData? SQLITE_ECEL_REF : 0)); if( bSeq ){ sqlite3VdbeAddOp2(v, OP_Sequence, pSort->iECursor, regBase+nExpr); } if( nPrefixReg==0 && nData>0 ){ sqlite3ExprCodeMove(pParse, regData, regBase+nExpr+bSeq, nData); } if( nOBSat>0 ){ int regPrevKey; /* The first nOBSat columns of the previous row */ int addrFirst; /* Address of the OP_IfNot opcode */ int addrJmp; /* Address of the OP_Jump opcode */ VdbeOp *pOp; /* Opcode that opens the sorter */ int nKey; /* Number of sorting key columns, including OP_Sequence */ KeyInfo *pKI; /* Original KeyInfo on the sorter table */ sqlite3VdbeAddOp3(v, OP_MakeRecord, regBase+nOBSat, nBase-nOBSat,regRecord); regPrevKey = pParse->nMem+1; pParse->nMem += pSort->nOBSat; nKey = nExpr - pSort->nOBSat + bSeq; if( bSeq ){ addrFirst = sqlite3VdbeAddOp1(v, OP_IfNot, regBase+nExpr); }else{ addrFirst = sqlite3VdbeAddOp1(v, OP_SequenceTest, pSort->iECursor); |
︙ | ︙ | |||
617 618 619 620 621 622 623 624 625 626 627 628 629 630 | sqlite3VdbeAddOp2(v, OP_IfNot, iLimit, pSort->labelDone); VdbeCoverage(v); } sqlite3VdbeJumpHere(v, addrFirst); sqlite3ExprCodeMove(pParse, regBase, regPrevKey, pSort->nOBSat); sqlite3VdbeJumpHere(v, addrJmp); } if( pSort->sortFlags & SORTFLAG_UseSorter ){ op = OP_SorterInsert; }else{ op = OP_IdxInsert; } sqlite3VdbeAddOp4Int(v, op, pSort->iECursor, regRecord, regBase+nOBSat, nBase-nOBSat); | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > < < < < < < < < < < < < < < < < < < < < < < < < < < < < | 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 | sqlite3VdbeAddOp2(v, OP_IfNot, iLimit, pSort->labelDone); VdbeCoverage(v); } sqlite3VdbeJumpHere(v, addrFirst); sqlite3ExprCodeMove(pParse, regBase, regPrevKey, pSort->nOBSat); sqlite3VdbeJumpHere(v, addrJmp); } if( iLimit ){ /* At this point the values for the new sorter entry are stored ** in an array of registers. They need to be composed into a record ** and inserted into the sorter if either (a) there are currently ** less than LIMIT+OFFSET items or (b) the new record is smaller than ** the largest record currently in the sorter. If (b) is true and there ** are already LIMIT+OFFSET items in the sorter, delete the largest ** entry before inserting the new one. This way there are never more ** than LIMIT+OFFSET items in the sorter. ** ** If the new record does not need to be inserted into the sorter, ** jump to the next iteration of the loop. Or, if the ** pSort->bOrderedInnerLoop flag is set to indicate that the inner ** loop delivers items in sorted order, jump to the next iteration ** of the outer loop. */ int iCsr = pSort->iECursor; int iJmp = sqlite3VdbeCurrentAddr(v)+5+(nOBSat<=0)+pSort->bOrderedInnerLoop; assert( pSort->bOrderedInnerLoop==0 || pSort->bOrderedInnerLoop==1 ); sqlite3VdbeAddOp2(v, OP_IfNotZero, iLimit, sqlite3VdbeCurrentAddr(v)+4); VdbeCoverage(v); sqlite3VdbeAddOp2(v, OP_Last, iCsr, 0); sqlite3VdbeAddOp4Int(v, OP_IdxLE, iCsr, iJmp, regBase+nOBSat, nExpr-nOBSat); VdbeCoverage(v); sqlite3VdbeAddOp1(v, OP_Delete, iCsr); } if( nOBSat<=0 ){ sqlite3VdbeAddOp3(v, OP_MakeRecord, regBase+nOBSat, nBase-nOBSat,regRecord); } if( pSort->sortFlags & SORTFLAG_UseSorter ){ op = OP_SorterInsert; }else{ op = OP_IdxInsert; } sqlite3VdbeAddOp4Int(v, op, pSort->iECursor, regRecord, regBase+nOBSat, nBase-nOBSat); } /* ** Add code to implement the OFFSET */ static void codeOffset( Vdbe *v, /* Generate code into this VM */ |
︙ | ︙ |