[Bjarte Aagnes]

.ster to JSON and JSON to .ster

[Bert Lloyd]

While in mata interactively:

1. allow selecting mata code from the do-file editor and Control-D to run the selected mata code

2. if I enter do dofile.do interactively or hit Control-D from the do-file editor, quietly end mata interactive mode and run the do-file without an error.

(I realize these might be a bit contradictory, even just one of them would be helpful if it's not feasible to add both. I don't really have a preference for one over the other.)

[Bert Lloyd]

.ster to JSON and JSON to .ster

.ster and .sters (from the excellent estwrite package contributed by the most excellent Ben Jann) are underappreciated IMHO.

It would be great if there were a way to make .ster (and .sters) files more human-readable. As I understand it they are binary files, they look pretty garbled if opened in a text editor. Being able to look at the contents would help with debugging, double-checking, etc.

(Perhaps there is already a way to view these, I would be very happy to be corrected if so.)


Jann, B. 2005. estwrite: Stata module to store estimation results on disk. Available from http://ideas.repec.org/c/boc/bocode/s450201.html.

[Silvia Gib]

I really hope that Stata 19 will let us import parquet files, they are everywhere now. The only solution I found is an user written command and only works with Linux.

[Fernando Furquim]

I'd like to see the xline and yline options in graphing commands support coordinates for the length of these lines when we don't want them to span the whole graph. It'd make it much easier than current workarounds

[Stephen Jenkins]

Fernando Furquim In the meantime, check out -graph twoway pci-. (Straighforward to use in my experience.)

[George Hoffman]

regarding graph combine: an 'overlay' option woulld be great to superimpose many plots into one set of axes.
the existing xcommon and ycommon options take care of scaling; there would be some legend and labeling logic to work through, but to get the plots on one axis set would be a great step forward for at least a few of us.
thanks for considering

[Bruce Weaver]

This wish concerns explicit inclusion of default options in code and explicit labeling of the option that was used in the corresponding output. I'll use -roctab- to illustrate, but I think this issue applies to many more commands. Here is some code:

use https://www.stata-press.com/data/r18/hanley, clear
foreach method in "" bamber hanley binomial {
    roctab disease rating, summary `method'
}

And here is the output it produces:

. use https://www.stata-press.com/data/r18/hanley, clear
(Tomographic images)

. foreach method in "" bamber hanley binomial {
  2.         roctab disease rating, summary `method'
  3. }

                      ROC                     Asymptotic normal  
           Obs       area     Std. err.      [95% conf. interval]
     ------------------------------------------------------------
           109     0.8932       0.0307        0.83295     0.95339

                      ROC      Bamber         Asymptotic normal  
           Obs       area     std. err.      [95% conf. interval]
     ------------------------------------------------------------
           109     0.8932       0.0306        0.83317     0.95317

                      ROC      Hanley         Asymptotic normal  
           Obs       area     std. err.      [95% conf. interval]
     ------------------------------------------------------------
           109     0.8932       0.0320        0.83047     0.95587

                      ROC                       Binomial exact   
           Obs       area     Std. err.      [95% conf. interval]
     ------------------------------------------------------------
           109     0.8932       0.0307        0.81559     0.94180

The documentation for roctab says:

By default, roctab calculates the standard error for the area under the curve by using an algorithm suggested by DeLong, DeLong, and Clarke-Pearson (1988) and asymptotic normal confidence intervals.

But there appears to be no way to explicitly include the Delong et al method as an option in the code. E.g., this causes an error:

roctab disease rating, summary delong

More importantly, IMO, I think that the SE should be labeled as the DeLong SE when the default option is used so that users do not have to go digging into the documentation to find out what the default is. I.e., I reckon the output should look like this:

. roctab disease rating, summary

                      ROC      DeLong         Asymptotic normal  
           Obs       area     Std. err.      [95% conf. interval]
     ------------------------------------------------------------
           109     0.8932       0.0307        0.83295     0.95339

As I said above, I suspect the same general idea applies to many other commands.