Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: FAQ visible on user-written ados and Stata version


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: FAQ visible on user-written ados and Stata version
Date   Thu, 5 Feb 2009 10:21:01 -0000

Sergiy makes two main points here. 

1. On StataCorp's update policy, he has spotted some sloppy wording. 

Of the two sentences he cites 

"If your Stata version differs by less than 1 from what is required,
you can update your Stata for free over the web. You can update for
free within an integer version, for example from Stata 9 to Stata 9.2
or from Stata 10 to Stata 10.1; just type update"

the first is ambiguous but is not intended to imply that because (10.1 -
9.2) < 1 you can -update- for free from 9.2 to 10.1. (In any case, that
would not be an -update- only, but an upgrade as well.) The second can
stand. 

We'll fix the wording. Thanks for the catch. 

2. On what Sergiy calls build numbers (which Stata internally knows by
another name), he is also right. The FAQ does not discuss this, nor was
it the intention to descend to that level of detail. 

There is always a judgement call over quite how much detail to include
in something like a FAQ. It's more important that it be clear to readers
than it be complete (or even that it be completely correct; sometimes
there are minor complications or exceptions that can be suppressed). The
FAQ is aimed at users asking simple questions about -version- and it
does not purport to tell the whole story. 

There is also always a judgement call about disclosing undocumented
features, for reasons Sergiy mentions and for reasons he does not
mention. 

Sergiy is right that users, and user programmers, need on occasion be
sensitive to different releases of the executable. This is all better
discussed in terms of visible details like born dates. In my experience,
this kind of detail is actually less puzzling than the issues discussed
in the FAQ. The careful programmer puts in a test for the user's c(born)
and if it is less than required issues an appropriate error message,
which is then transparent (e.g. you need Stata of at least 28 March 2009
to run -foobar-). Also, I suspect most people concerned with this kind
of detail have, like Sergiy, already worked it for themselves. 

I'll hear advice about this, especially from StataCorp as despite my
name being on it it's their FAQ, but my own judgement call is that in
some ways the FAQ is already too long and complicated for the majority
of its intended readers. 

Nick 
n.j.cox@durham.ac.uk 

Sergiy Radyakin

I would like to thank the authors of the FAQ and hope that many Stata
users will benefit from the information they can find there. But I'd
like to know if the updates policy of Stata has changed recently? is
the document entirely accurate? In particular I mean this statement:
"If your Stata version differs by less than 1 from what is required,
you can update your Stata for free over the web. You can update for
free within an integer version, for example from Stata 9 to Stata 9.2
or from Stata 10 to Stata 10.1; just type update"

The update policy known to me was: "within one major version, minor
versions are free".

Is StataCorp already flooded with requests to upgrade Stata 9.2 to
Stata 10.1 fo free? (10.1-9.2) = 0.9<1or is there a small-font to it?

Another thing which seems to be missing from this FAQ is the
(undocumented) build number. Within one version e.g. version 10.0
there could be several public releases with each subsequent release
fixing some bugs found during that time. This does not change the
version-minor (10 is major, 0 is minor), but can be an additional hint
why the program does not work on a particular machine, though works on
another, with Stata reporting the same version. For example, while
working with the graphics commands I've experienced a problem with
Stata 10, which I didn't have in Stata 9. I didn't report it since the
features I am using are not documented, and StataCorp does not
guarantee that they will work, but perhaps someone else did report it,
or they were removed by mistake. It was changed later, and now I know
that my command works in the latest 9.2 and in 10.0.539 or later, but
it does not work properly in 10.0.526d.

While the build number is not widely publicized, the born-date is more
widely known. It is displayed after -about- and also after -update
query-.

On Wed, Feb 4, 2009 at 1:33 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:
> On 24 January David Elliott in
>
> <http://www.stata.com/statalist/archive/2009-01/msg00914.html>
>
> suggested a FAQ on user-written ados and Stata version. StataCorp
kindly
> agreed to host such an FAQ and it is now visible at
>
> <http://www.stata.com/support/faqs/lang/version2.html>
>
> The FAQ as posted benefits from discussions on the list by David, Ben
> Jann and Richard Williams and from intervention off the list by
William
> Gould of StataCorp. The nice presentation is due to Annette Fett of
> StataCorp.
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index