[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: Choosing between SSC & SJ for ado files
I'll put a hat on an Editor of the SJ (Stata Journal).
I am also an original author of -ssc- and a contributor
to SSC but Kit Baum speaks definitively on that.
The Stata Journal appears quarterly and so no item
associated with the SJ is publicly updated more frequently
than that. (Occasionally we make quick quiet fixes for
the greater good, but the principle of no publication
without publicity is good in many ways.)
SSC can be updated at Kit's discretion, modulo his
sleeping patterns, leisure time, academic commitments and
vacations. In my experience the median time between
a submission and an update is much less than a working day,
which indicates Kit's efficiency and dedication.
So, SSC can be updated much more frequently than the SJ.
It follows that where two versions exist it is often
the case that the version of SSC is the more recent.
But to that should be added other comments.
The first is, Do check the code!
This is easier than you might think, even if you are
not (consciously) a Stata programmer. Thus let's
test this example. If I type
. ssc type freduse.ado
I get to see a listing of the main program that includes
*! version 2.0.0 19Sep2005
Non-programmers: You are looking for a line that starts
and if you see no such line you should be worried,
as the programmer is not following standard Stata conventions.
That doesn't prove that they are incompetent, but in
practice it is a good warning in most cases.
. search freduse
I get to see details of where -freduse- is stored on the
Stata Journal website. Without even installing it I can
. type http://www.stata-journal.com/software/sj6-3/st0110/freduse.ado
and I get to see the same file. The principle here is
simple concatenation. I use the Stata command -type- and
the information on -freduse- includes all the ingredients for
a specification of where that file is. (Naturally, this does depend
on (a) an up-to-date Stata, as -update- updates the .key files
containing information on the SJ, and (b) an internet connection.)
Now, this package is more
complicated than most and there are other program files
or help files that may not be exactly the same. However, in
my experience peeking at the files like this resolves the question very
quickly in most cases.
The second is to note that there has been quite a lot of discussion
behind the scenes about this situation. One extreme view, which
has a lot to be said for it, is that programmers should maintain
exactly one source for their files. It undoubtedly is at least
a little confusing when there are two (or even more) sources
for a program, even setting aside the situation that some kind-hearted
programmers make available versions of their programs for both
the present version of Stata and for previous versions.
The Stata Journal follows a variety of procedures to maintain its own
standards, but we at the Stata Journal have no desire, and no right,
to lay down the law on how and where user-programmers should be
disseminating their programs. So we never forbid or even advise
against dissemination on SSC or elsewhere. Nevertheless I don't think that
anyone will be offended by a remark that there being just one source does
simplify things for the end-user.
Michael S. Hanson
> I am interested in installing the -freduse- command
> that was just
> announced in SJ 6-3. (This command allows one to load macroeconomic
> data directly into Stata from the Federal Reserve Economic Database
> (FRED) at the St. Louis Fed.) -findit freduse- returns two sources
> for the corresponding package:
> 2 packages found (Stata Journal and STB listed first)
> st0110 from http://www.stata-journal.com/software/sj6-3
> SJ6-3 st0110. Importing Federal Reserve economic data /
> Importing Federal
> Reserve economic data / by David M. Drukker, StataCorp / Support:
> email@example.com / After installation, type help freduse
> freduse from http://fmwww.bc.edu/RePEc/bocode/f
> 'FREDUSE': module to Import FRED (Federal Reserve Economic
> Database) data
> / The FRED repository at
> http://research.stlouisfed.org/fred2/ /
> over 3,000 U.S. economic time series. Each time series / is
> stored in a
> separate file that also contains a string-date / variable and
> header with
> Based on this message from Kit Baum,
> do I understand correctly that installing from SSC (that is, RePEc)
> is preferable if a package is updated more frequently than once a
> quarter and one plans to use -adoupdate- to manage installed
> packages? Thanks for any clarification on this question.
* For searches and help try: