[Matthew White]

Another Java wish: it would be great to be able to query whether Stata has or could load a specified class.

I have the following scenario in mind. Let's say I write a Java helper class named Helper and add it to SSC. I then write a new class named MyClass that uses Helper, adding that to SSC. The MyClass package comes with the Stata wrapper program -myclass- so that the user doesn't have to interact with Java in any way.

I don't want to distribute Helper with MyClass, because I don't want to have to update the MyClass package every time I update Helper. Yet if the user downloads MyClass without Helper and tries to run -myclass-, a ClassNotFoundException will be thrown and the user will see a Java stack trace. An error is inevitable here, but I'd like to issue a error message more informative to a non-Java user, something like "SSC package helper required; to install, type ssc install helper". If MyClass could query from Stata whether Helper is loaded, I could do that. I could have MyClass display an error, then return to Stata with a nonzero return code before the exception is thrown.

[Brian Quistorff]

My wish list includes mainly items for reproducing research:

1. You should be able to download (with -net-/-ssc-) old versions of modules easily. This would probably require the package format to include versions explicitly (e.g. in the pkg file not just as *! comments in the ados) and repositories (SSC) to save old versions.
2. Primary output files (dtas and gphs) should be able to be reproduced byte-for-byte. Primarily this requires being able to zero-out the timestamps and zero-out any junk padding.
3. Make PDF and PNG exporting of figures available on console Unix.
4. Shell commands should work in Windows batch-mode.
5. Built-in commands should almost always return data in s()/r(), but some return just to the screen.
6. Also, allow the Windows do-editor to automatically word wrap (this is a main reason why people I know use other editors).

[Clyde Schechter]

I would like to see commands that export data to other formats (I'm thinking of -export delimited- and -export excel-) to include a -usedisplayformat- option that applies the existing display formats to each variable affected in the target file.

[David Muhlestein]

It would also be nice if Stata could increase their limits across the board (number of variables, length of labels, number of label values,etc.). Long strings in Stata 13 were a great addition, but it should be expanded elsewhere. I understand that there are reasons for some of the limits, but being able to override them would be wonderful.

[ben earnhart]

Be nice if -alpha- could work off a matrix (like factormat can), or had an option to use polychorics. For that matter, might bring polychoric into the officially supported commands instead of keeping it as an add-on.

[Kieran McCaul]

Part of what my profile.do does is set my ado-path, which contains about 75 directories, as my ado-files are scattered across project directories and Git repositories. It also sets my PERSONAL system directory outside the default C:\ado\personal: I like keeping PERSONAL on Dropbox in order to facilitate ado-file consistency across machines. Yet all my calls to the adopath and sysdir commands seem to be processed after the JRE is loaded fairly early on, so javacall looks only in C:\ado for Java files. Again, a command to reload the JRE with the current ado-path would address this.

I was having this problem when I moved my ado-path to dropbox. When I sent a do-file to Stata from my editor, if Stata was not already running, the ADOPATH would not be set correctly. So I moved my sysdir and adopath commands to sysprofile.do which, if it exists, is run by Stata before profile.do.

[Nathan E. Fosse]

I would like Stata 14 to have a suite of latent class analysis (LCA) or latent structure models. Presently, the most popular software are Mplus, Latent GOLD, and the flexmix() and poLCA() packages in R. This would be great for Stata's market share, as these analyses are ubiquitious in market research.

In Stata, fmm by Partha Deb and the notable gllamm ado packages provide aspects of these, but are somewhat more limited than Mplus or Latent GOLD. (gllamm in particular requires more patience than I can muster).

Perhaps there is a way to run LCA models in Stata using gsem; I haven't personally gotten around to using the sem commands yet (although Acock's book sits on my bookshelf).

Does anyone know if gsem possibly provides latent class analyses?

Cheers,
- Nate
P.S. In the Stata environment, if you're going with Mplus, I definitely recommend Rich Jones' and collagues' parsing of Mplus syntax, runmplus (link below). By far, even within R, it's the best means of integrating Mplus in any general stats package!

[Mark Schaffer]

I'd like to see a nooutput option appear as standard on estimation etc. commands.

Why isn't quietly or capture good enough? Because these suppress all the output, including any detailed error or warning messages. For example, say I call regress as an intermediate step in a do or ado file. I don't want the regression output reported. But if regress finds collinearities and drops some regressors, then I want it to report the corresponding "note: x omitted because of collinearity" but not the main output.

Incidentally, this is possible with regress by using the noheader and notable options together. But regress is exceptional in this regard.

[Konrad Zdeb]

I would like to see commands that export data to other formats (I'm thinking of -export delimited- and -export excel-) to include a -usedisplayformat- option that applies the existing display formats to each variable affected in the target file.

How about dataout (SSC)?

[Marcos Almeida]

I'd be very happy with power and sample size estimations for mixed models. Also, a complete menu concerning metanalysis. Wishful thinking?

[Lui Yiu Lim]

I hope that stata could make improvement in the postfile command. Now it saves the output when all iterations are finished.
So, when you perform a 10-time iteration, if there is problem in the 9th time, all the previous 8-time iteration's data are gone.

I hope that stata will let user to set "a save point", or set how frequent stata is to updata the file.

[Stephen Jenkins]

Re post #176: this functionality already exists. Please see the every(#) to postfile

[László Sándor]

This is the future of Stata too, I hope: https://www.youtube.com/watch?v=iPJJmRxxkZ0

[Clyde Schechter]

Somewhat related to #176: when running -simulate- and saving the default _b estimates, the file comes out with variable names like _sim_1, _sim_2, etc., and the variable labels correspond to the names of the coefficients that appear in the e(b) column names. And that's great. But if you're doing a very long simulate, and saving with the -every()- option and need to stop the simulation before the planned number of replications, the file does not have the variables labeled. Yes, you can recover them from e(b) but it's a (minor) nuisance. It would be nice if -simulate- applied the variable labels when it first creates the file, rather than only after the last replication.

[Arne Risa Hole]

Re post #172: see also the excellent lclogit package by Pacifico and Yoo and the associated SJ paper: http://www.stata-journal.com/article...article=st0312