|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: SSC activity, August 2006
Austin and others,
I apologize to Statalist contributors for being hassled by questions
about the outdated -outreg- program. I have long admired the dedication
of Statalist contributors, and don't like to get in their way.
Contrary to the impression I get by perusing the Statalist archives,
however, -outreg- it has never been "unsupported". I get weekly if not
daily technical support emails, many of them forwarded by Statacorp tech
support. I respond to them all with rare exceptions (a sore point with
my wife). I have always been reachable at the email addresses in -outreg-.
What I have not had time to do is participate in Statalist. Six years
ago I left academia to raise my young boys and design houses in Maine
while my wife went back to medical school. During those years I also
did not have time for new Stata program development.
-outreg- and its more up-to-date counterparts -outreg2- and -estout-
face highly unusual expectations for ado programs. Whenever Stata
estimation commands or statistics evolve, users expect the new
capabilities to be supported by -outreg-. -outreg- is about a close to a
bug-free program as software gets after years of use by many thousands
of users, but I have not been able devote the time to adapt it to new
Stata capabilities until recently.
I have now returned to economic research and teaching, and in the last
couple of months I have been rewriting -outreg- in Mata, a wonderful
language. I have ambitious plans - to create a back end for -outreg-
that will allow users to fully format any Stata output, not just
estimation output (for example summary statistics and -tabulate-
output), and merge these tables together. Mata for the first time
provides the appropriate data structure, the string matrix.
For the time being, I would appreciate it if you could recommend to
people asking -outreg- questions that they contact me directly (after
you explain that -outreg2- and -estout- are better :). I can always be
reached at the address in -outreg-: [email protected]. If I do
not complete my new version of -outreg- in the next several months, I
will agree to have all pointers set to -outreg2- and -estout-.
Take care,
John Gallup
Austin Nichols wrote:
Friedrich--
The proposition that there is not a "never-ending stream of messages"
complaining about some behavior of -outreg- is unverifiable
empirically, I suppose, since we will never know if we have seen the
end of them. Perhaps the distinction between "continuous" (i.e.
without interruption; no such claim was made) versus "continual" (i.e.
continuing) messages is relevant. I am confident most frequent
Statalist contributors recognize my point, however, and are familiar
with both the complaints and the replies of the form "quit using
-outreg- and use -outreg2- or -estout-" e.g.
http://www.stata.com/statalist/archive/2006-08/msg00477.html
http://www.stata.com/statalist/archive/2006-06/msg00740.html
My proposal is not that we unilaterally remove -outreg- but that we
ask its author to support it or allow its name to be used for newer
updates, which would be particularly valuable if the author of
-outreg2- committed to ensuring backward compatibility with -outreg-.
I am emphatically not denigrating the contribution of John Gallup, but
that contribution has not been updated since May 2002, and as a result
now often appears broken, or at least inadequate to some hopeful
user's needs, in a way that -estout- -estout1- and -outreg2- do not.
If -outreg- could be renamed -outreg6- (reflecting its "version 6"
declaration) then -outreg2- could be renamed -outreg- with little
fanfare. In the absence of this remedy, Gallup should at least add a
line to the help file for -outreg- clarifying that it is no longer
supported, and interested users should get -estout- -estout1- or
-outreg2- from SSC.
As to the general principle of never changing a file on SSC in a way
that would "break the code" in existing do-files, that is clearly a
counterfactual desideratum. That is, as much as we might like as a
general rule that SSC programs should remain forever backwards
compatible for our convenience, they do not. Programs on SSC are
regularly updated, and do not always preserve prior behavior. Some
programs are updated to make use of new functionality in official
Stata, and the prior version is sometimes renamed, or simply
disappears.
This message is not influenced by personal bias. I use none of these
programs, but rather the -file- and -local- commands, and nested
-foreach- loops.
On 9/3/06, Friedrich Huebler <[email protected]> wrote:
Sorry, but I disagree with this proposal for several reasons. First,
there is no "never-ending stream of messages beginning "I have a
problem running outreg."" Second, the continuing popularity of
-outreg- indicates that the program fulfills the needs of most users.
Third, letting -outreg2- become -outreg- may break the code in
thousands of existing do-files. Fourth, I don't think it is fair to
ask John Gallup to withdraw his valuable contribution to Stata.
This message is not influenced by personal bias. I use -estout-.
Friedrich Huebler
--- Austin Nichols <[email protected]> wrote:
> Could we not just ask Gallup <[email protected]> to withdraw
> -outreg- and let -outreg2- become -outreg- on SSC? It would save
> all of us a never-ending stream of messages beginning "I have a
> problem running outreg."
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
