Re: Is replacing transactions with CTE a good idea?

Ah, I see what you mean. You still have to wrap a CTE inside a transaction to specify the isolation level? By default, queries in a CTE run with the read committed isolation level?

> On Apr 1, 2021, at 11:10 PM, Dave Cramer <davecramer(at)postgres(dot)rocks> wrote:
>
> 
>
>
>> On Thu, 1 Apr 2021 at 11:09, Glen Huang <heyhgl(at)gmail(dot)com> wrote:
>> No, but are they equivalent to serializable transactions?
>
> No, they are not.
>
>
>
> Dave Cramer
> www.postgres.rocks
>>
>>>> On Apr 1, 2021, at 11:04 PM, Dave Cramer <davecramer(at)postgres(dot)rocks> wrote:
>>>>
>>> 
>>>
>>>
>>>
>>>> On Thu, 1 Apr 2021 at 10:50, Glen Huang <heyhgl(at)gmail(dot)com> wrote:
>>>> Hi all,
>>>>
>>>> From application’s standpoint, it seems using CTE saves a lot work. You no longer need to parse values out only to pass them back in, and only one round-trip to the db server.
>>>>
>>>> If I’m not wrong, CTE is equivalent to serializable transactions? So I guess the downsize is that quarries can’t be run in parallel?
>>>
>>> I do not think a CTE changes the isolation level.
>>>>
>>>> If I decide to replace all my transaction code with CTE, will I shoot myself in the foot down the road?
>>>
>>>
>>> Dave Cramer
>>> www.postgres.rocks