Re: [Maria-developers] Questions about AS OF (possibly bugs found)

2018-04-11 Thread Sergei Golubchik
Hi, Aleksey! On Apr 11, Aleksey Midenkov wrote: > Hi Sergei, Alexander! > > Maybe it's better to remove TIMESTAMP/TRANSACTION auto-detection as it > requires too much code to make it properly? And make TIMESTAMP by > default. I'd rather move it down when all items are fixed. If possible.

Re: [Maria-developers] Questions about AS OF (possibly bugs found)

2018-04-11 Thread Aleksey Midenkov
Hi Sergei, Alexander! Maybe it's better to remove TIMESTAMP/TRANSACTION auto-detection as it requires too much code to make it properly? And make TIMESTAMP by default. On Mon, Apr 9, 2018 at 5:49 PM, Sergei Golubchik wrote: > Hi, Alexander! > > On Apr 09, Alexander Barkov

Re: [Maria-developers] Questions about AS OF (possibly bugs found)

2018-04-09 Thread Sergei Golubchik
Hi, Alexander! On Apr 09, Alexander Barkov wrote: > > Note, history_point adds around 20 shift/reduce conflicts. > I'd like to resolve them. good :) > > Might've been better to look at cmp_type() though. > > A new method in Type_handler would be even better. okay > >> void

Re: [Maria-developers] Questions about AS OF (possibly bugs found)

2018-04-09 Thread Alexander Barkov
Hello Sergei, On 04/09/2018 01:09 PM, Sergei Golubchik wrote: > Hi, Alexander! > > On Apr 09, Alexander Barkov wrote: >> Hi Sergei, >> >> I'm looking at this use rule in sql_yacc.yy: >> >> history_point: >> temporal_literal >> { $$= Vers_history_point(VERS_TIMESTAMP, $1); }

[Maria-developers] Questions about AS OF (possibly bugs found)

2018-04-09 Thread Alexander Barkov
Hi Sergei, I'm looking at this use rule in sql_yacc.yy: history_point: temporal_literal { $$= Vers_history_point(VERS_TIMESTAMP, $1); } | function_call_keyword_timestamp { $$= Vers_history_point(VERS_TIMESTAMP, $1);