Add more data types to the column header - and improve the display of lists

[N-thony]

@rdstern @dannyparsons could we now have the possibility to make datetime?
I have been looking this link https://stackoverflow.com/questions/44970127/remove-timezone-during-posixlt-conversion-in-r
@lilyclements what do you think?
Probably add an option in Year - Month -Day, a checkbox when it is checked it open a groupbox containing hour,min,sec input controls where we can fill and make a datetime and when it is unchecked we make date simply

[N-thony]

I'm not sure if we need this since we can make datetime through the Calculator Dailogue

[lilyclements]

We can run any R function through the calculator dialog so I don't understand why this is a problem?
In R-Instat we can give a front end to the code. Do you think this is something this is useful to have in a front end? I think it sounds sensible, and in the PLH work there are variables that are date-time. But what do you think? I'm not sure of the reasons why this hasn't been previously implemented.

[rdstern]

For now I am happy to use the calculator, but later I suggest it will be good to also have a dialogue. But also, for now, we currently have type (D) for all these types of data. I feel the types would be clearer in R-Instat if we kept (D) for a date variable and had (DT) for date-time and (T) for time. I have sometimes found it confusing when some current type (D) variables don't work, as I expect, because they are (DT).
Perhaps @dannyparsons can confirm or deny?

[dannyparsons]

Roger's suggestions sounds sensible to me.

[rdstern]

@N-thony version 0.7.6 is holding up well in the important India (IMD) workshop that I am finishing today. The only 2 bugs/inconveniencies have been a bug in the use of the PICSA rainfall graphs (Vitalis is looking at that) and the confusion here about date-time variables.
In the data I have sent you, the first sheet had a (current) date variable. And it looks like an ordinary date too.
a) If you examine the metadata the variable called ...1(D) appears as a class POSITct.POSITt rather than date.
b) If you use the Describe > One Variable > Summarise on it, then the summary includes the time element.
It is particularly confusing, because it presents with just the date part.

a) I suggest it be given the type (D.T) instead. (That's like an ordered factor is (O.F).
b) We may stop there, or also consider time variables as of type (T) at the same time.

Thanks

[rdstern]

@N-thony can you do this? It would be good to avoid confusion. Could you do a) above? Or should someone else , e.g. @dannyparsons or @lilyclements do this?
I ask because - at the same time - I have another type of variable to specify which I mention in another issue. But it would be good to do it at the same time.

The other type is different. An example is File > New Data Frame > Lists > Words/Literature > Shakespeare/sonnets. The sonnets come as a list. They don't have anything attached. Could we have (L) for list please - at the same time

[lilyclements]

Could we have (L) for list please - at the same time

@rdstern we currently use (L) for logical. Is there a different letter you can think we should use here?

[rdstern]

I am wondering if we have the general (S) for structure. As you say (L) is already in use.

[N-thony]

I am wondering if we have the general (S) for structure. As you say (L) is already in use.

@rdstern (S) is already used for structure. What if we keel (L) for List and (B) for Logical?

[rdstern]

@N-thony the idea is that S for structure is general that it has a special structure. So our circular data is already of type S. So I suggest that Lists should also be of type S at least for now. In addition I have been working on the Integer keyboard. This produces variables of class bigz or bigq, see below: They should also have type S.

I would like to return to the situation that if there is no structure given, then the variable is numeric. Everything else has a letter after it.

[rdstern]

@N-thony and possibly @lilyclements, can we add either S type for lists as a structure, or perhaps (LT) so we detect lists. And could we please display lists in reogrid so they are the same as in R?

Here is an example from a new data frame with 1 to 100 in x1 and then a list in x2 from Prepare > Calculator > Integer > Divisors.

You can see that in R it is a nice simple list, but it looks much more complicated in R-Instat. I wonder why? And I wonder too about the extra variable in the metadata. Do we need that?

This display in reogrid creates big problems if you try to edit the data. Then it presumably ceases to be a list, and so gives errors.

[N-thony]

@N-thony and possibly @lilyclements, can we add either S type for lists as a structure, or perhaps (LT) so we detect lists. And could we please display lists in reogrid so they are the same as in R?

Here is an example from a new data frame with 1 to 100 in x1 and then a list in x2 from Prepare > Calculator > Integer > Divisors.

You can see that in R it is a nice simple list, but it looks much more complicated in R-Instat. I wonder why? And I wonder too about the extra variable in the metadata. Do we need that?

This display in reogrid creates big problems if you try to edit the data. Then it presumably ceases to be a list, and so gives errors.

@rdstern this issue of displaying the lists is related to #7947 and it seems to be fixed in #8015

[rdstern]

@N-thony the variables of type bigz and bigq are now supposed to be of type (S) - this was merged in #7894.

But they are not any more? This has confused @rachelkg who has been producing them.

[N-thony]

@N-thony the variables of type bigz and bigq are now supposed to be of type (S) - this was merged in #7894.

But they are not any more? This has confused @rachelkg who has been producing them.

@rachelkg which version of R-Instat are you working on? This was done in PR #8015 and merged.