reworked beqs()

[eddelbuettel]

Please take a look as time permits.

This is similar to what @armstrtw does with LazyFrame but more direct / brutish: we do one pass over the (first) data type, allocate and assign typed Rcpp vector accordingly and then do loop over the data and do the switcheroo on type.

So far only character and float64 tested. Adding Date(time) should be straightforward.

Ohh, and I never have to say SEXP or PROTECT :)

[csrvermaak]

Great stuff @eddelbuettel, thank you.

[eddelbuettel]

Let me know if you have issues getting it built or not.

I think the patch should be fine as is, but we need to do some more tweaking. Eg right now I do not add pitdate back in as it may be empty. We may just want to always add it as the first column, with current date as default.

It would also be nice to see some screens returning date or datetime (if they exist).

[csrvermaak]

@eddelbuettel - It builds 100%, no problems. :)

I agree, the pitdate should default to today() if empty, but should always be included, otherwise the data is a little "lost", not being anchored in time.

If you need a result that has a date/datetime column, I can construct an EQS example and share with you. Would need your Bloomberg ID so I can MSG it?

[eddelbuettel]

So you are a gitmaster now :) Nice. 'git-bash' on Windows, or their fancy 'github desktop' ?

So how about if we always return the date as the first column? Ie instead of

if (!is.null(date)) res <- data.frame(pitDate=date, res)

in beqs.R, we always set it?

I'll get you the account handle once I get to the office. Likely Kevin Saunders @ Ketchum Trading but I'll check.

[csrvermaak]

github desktop, it's awesome! Not a gitmaster yet... still earning my stripes with a few hiccups and bumps, but loving the learning curve. I can absolutely see the advantages over svn, and very glad I switched!

Returning the pitdate as the first column would be the ideal solution.

[wmorgan85]

I can check eqs public samples when I get back to my desk shortly for date columns. Will post in a few.