[Nick Cox]

if I change the name of a file somewhere, it changes everywhere I've referenced it automatically

I really don't want that. Really, really, really not. But no matter: if you want that, put the filename in a global macro. That's what they are for.

Am I misunderstanding?

[wbuchanan]

Mansi Jain
It might be useful to distinguish between what you would like StataCorp to implement and existing capabilities.

[Mansi Jain]

I'd really like markdown on do-files -- would be super helpful to keep code organized. I'd also love bidirectional linking -- for instance, if I change the name of a file somewhere, it changes everywhere I've referenced it automatically.

[Chen Samulsion]

The new popped up thread reminds me of a longlasting wish. Stata should give users choice to dock Command Window in main interface or dock Dofile Editor in main interface. Users like me may want to see program results immediately in Result Window. https://www.statalist.org/forums/for...do-file-editor

[Bruce Weaver]

Imagine a two-way ANOVA model with factors A, B and their interaction. It appears that the best/easiest way to get Type II SS for this model is to do the following:

anova B A A#B, sequential // For A, use SS, MS and F from this model
anova A B A#B, sequential // For B, use SS, MS and F from this model

SS(AxB) and SS(error) can be taken from either model.

Compared to other stats packages where one can specify Type I, II, III or IV, this seems very clunky. I would like to see options for Type II and Type IV SS added to -anova-. (Personally, for consistency with other packages and resources, I would like to see the Type I, II, III and IV labels used too rather than partial, sequential etc.. I understand that others may take a different view.)

Some relevant discussions:

• https://www.statalist.org/forums/for...-anova-command
• https://www.statalist.org/forums/for...es-stata-and-r

[Andrew Wade]

Amending Putexcel, to allow it to use a number, in place of a letter for an Excel column.
This can currently be achieved using the ado Excelcol.
But it seems that including this feature within Putexcel would make sense.

Regards,

Andrew

[Dave Airey]

A command for comparing Pearson or Spearman independent or dependent correlations, with or without replication would be nice.

https://journals.plos.org/plosone/ar...l.pone.0121945

https://www.omicsonline.org/open-acc....php?aid=54592

-Dave

[Jean-Claude Arbaut]

As mentioned in this thread, add variable autocompletion to dialog boxes. That is, type a few letters of a variable, and then the drop down list only shows variables that match.
For cases where several variables can be entered (dependent variables in a regression for instance), maybe do it like in the command line: if the cursor is on a space show the full list, instead filter.

It would also be nice to update the drop down list as one types, but this is maybe more difficult to implement.

[Clyde Schechter]

Yes, from my perspective, the most important new feature of version 16 is frames. For me, that alone is worth the price of the upgrade. I will be using them extensively; in fact, I already am.

[Joseph Coveney]

Clyde Schechter
I had no idea about that and will definitely need to spend a bit more time checking out the frames functionality.

I recommend looking at frames to all users of the new release.

Capabilities like this are what I was alluding to at the end of #24.

[wbuchanan]

Clyde Schechter
I had no idea about that and will definitely need to spend a bit more time checking out the frames functionality.

[wbuchanan]

I didn't mention this previously, but I think adding a file watcher to the project manager would be a nice addition. Basically instead of having to add every subdirectory in a project or add all the individual files, the project manager file could either watch for new directories/files in the same location where it is located or would watch for new files in directories that were added to the project. It would save a bit of time when the projects being worked on contain a lot of files or when lots of files are generated as output.

[Clyde Schechter]

Also with regard to #18, you can now accomplish this by putting the two sets in different frames, -frlink- them (because -frlink- allows different names for the linkage, and then use -frget- to build up the combined data set.

[Rasool Baloch]

I will suggest on #18 to Mark Davis, please try to -ssc install mmerge-. You can merge data sets with different names of key variables and also rename variables from using dataset as well.

Regards,
Rasool Bux

[wbuchanan]

1. Native JDBC support
2. Allow mapping variable values to aesthetic properties in graphs (e.g., setting the color of scatter plot points based on a third variable instead of multiple calls to scatter).
3. SQL like syntax for intersections and unions of data (e.g., merges and appends).
4. Similar to what Richard Williams mentioned, I think performance enhancements under the hood could be great and would extend it to multi-level models and graphics too.