[sqlite] Lemon reduce action merge bug

2016-04-30 Thread Richard Hipp
On 4/30/16, Kelvin Sherlock wrote: > > Here are a couple more edge cases I noticed: > Thanks for the deep-diving on Lemon! All changes are now checked in. For passive listeners - none of the bugs that Kelvin has found in Lemon have any impact on SQLite. They only come up when using Lemon on ot

[sqlite] Lemon reduce action merge bug

2016-04-30 Thread Kelvin Sherlock
> On Apr 29, 2016, at 7:29 AM, Richard Hipp wrote: > > On 4/28/16, Kelvin Sherlock wrote: >> I believe the lemon reduce action optimizer needs to compare the codePrefix >> and codeSuffix. > > Thanks for the bug report. A fix has now been checked in. > > -- > D. Richard Hipp > drh at sqlite.

[sqlite] Lemon reduce action merge bug

2016-04-29 Thread Richard Hipp
On 4/28/16, Kelvin Sherlock wrote: > I believe the lemon reduce action optimizer needs to compare the codePrefix > and codeSuffix. Thanks for the bug report. A fix has now been checked in. -- D. Richard Hipp drh at sqlite.org

[sqlite] Lemon reduce action merge bug

2016-04-28 Thread Kelvin Sherlock
I believe the lemon reduce action optimizer needs to compare the codePrefix and codeSuffix. Consider this example: 8< %type malloc {void *} %destructor malloc {free($$)} %type contrived_example {int} bug_report ::= contrived_example. contrived_example(LHS) ::= INT INT INT. { LHS = 0; }