Support COUNT(*) for Aggregate COUNT function

[bobbai00]

Currently for aggregate operator, when selecting COUNT as the aggregation function, an attribute must be given. But in some use cases, there are no suitable attribute to select (e.g. there is no "primary key alike" attribute but users may want to count the occurance).

So a * attribute to support COUNT(*) should be useful.

[yunyad]

We decided not to proceed with it for now for several reasons:

1. The aggregate functionality includes many other options besides count, so it seems unreasonable to add it there. (Formly limitation)
2. Since this is not urgent, we can find a better position for this functionality.

[Yicong-Huang]

@chenlica @zuozhiw @aglinxinyuan @Yicong-Huang let's have an offline discussion.

[tanishqgandhi1908]

/take

[Yicong-Huang]

@aglinxinyuan @chenlica let's revisit this issue.

[chenlica]

@Yicong-Huang Can you remind us what we discussed before? I don't see a discussion result here.

@tanishqgandhi1908 Please put the effort on hold for now.

[chenlica]

Just saw the PR #5896 and the discussion there. Will continue the discussion there.

[Yicong-Huang]

Those are two different discussions. here we want to know if we have a need for this function. our discussion history is on this issue. In the PR we discuss how to implement the function.

[zuozhiw]

if we have a need: sure why not have count(*) it's virtually in every db system, maybe count() is also fine and it's equal to count()