Re: Simplify set of flags used by MyXactFlags

2019-01-25 Thread Michael Paquier
On Fri, Jan 25, 2019 at 12:30:39PM -0300, Alvaro Herrera wrote: > I am, except that the change of "table" to "object" in xact.c line 2262 > makes the new paragraph read a bit weird -- it's now saying "if we've > added a temp object ... Other objects have the same problem". Doesn't > make sense.

Re: Simplify set of flags used by MyXactFlags

2019-01-25 Thread Alvaro Herrera
On 2019-Jan-25, Michael Paquier wrote: > On Thu, Jan 24, 2019 at 08:56:05AM -0300, Alvaro Herrera wrote: > > Uh, I didn't think it was necessary nor desirable to rename the flag, > > only the user-visible message. > > Oh, OK. I have overstated your comment then. Are you fine with the > attached

Re: Simplify set of flags used by MyXactFlags

2019-01-25 Thread Michael Paquier
On Thu, Jan 24, 2019 at 08:56:05AM -0300, Alvaro Herrera wrote: > Uh, I didn't think it was necessary nor desirable to rename the flag, > only the user-visible message. Oh, OK. I have overstated your comment then. Are you fine with the attached instead? The flag name remains the same, and the c

Re: Simplify set of flags used by MyXactFlags

2019-01-24 Thread Alvaro Herrera
On 2019-Jan-22, Michael Paquier wrote: > On Mon, Jan 21, 2019 at 02:58:39PM -0300, Alvaro Herrera wrote: > > "... operated on temp namespace" doesn't look good; seems to me to be > > missing an article, for one thing, but really I'm not sure that > > 'namespace' is the term to be using here. I'd

Re: Simplify set of flags used by MyXactFlags

2019-01-21 Thread Michael Paquier
On Mon, Jan 21, 2019 at 02:58:39PM -0300, Alvaro Herrera wrote: > "... operated on temp namespace" doesn't look good; seems to me to be > missing an article, for one thing, but really I'm not sure that > 'namespace' is the term to be using here. I'd say "... operated on > temporary objects" instea

Re: Simplify set of flags used by MyXactFlags

2019-01-21 Thread Alvaro Herrera
On 2019-Jan-18, Michael Paquier wrote: > Keeping both messages makes the error handling at PREPARE time perhaps > a bit cleaner to make the difference about schema-level access or > table-level access, still I'd rather simplify the code and just only > keep the schema-level change as something we

Simplify set of flags used by MyXactFlags

2019-01-17 Thread Michael Paquier
Hi all, c5660e0 has introduced a new flag for MyXactFlags to restrict two-phase commit from working with temporary objects, and as a matter of fact XACT_FLAGS_ACCESSEDTEMPREL has been kept around to keep the error handling message compatible with past versions, still it is weird to keep both ACCES