laszlo kardos
>
> i recently had to dive deeper into low level graphing than
> i ever thought i
> would, and the experience made me question the
> unquestionable, namely that
> stata is by far the greatest piece of software, over all
> categories, i ever
> had the chance to pirate (kidding). here's the list of the
> most important
> issues.
>
< snip >
>
> i understand that gph is under replacement with something
> better. the latter
> two are mainly convenience issues, but for the data(xy) to
> window(rc)
> conversion to work with pinsharp precision is still a
> minimum requirement in
> my book.
>
> i would be grateful for workarounds, ideas, and development
> plans/wish lists
> from users and statacorp people alike.
Stata Corp have made it clear that a substantial graphics
project has been underway for some time, and so we
all hope that it will be included in the next release.
The implication of this is really that tinkering with
-gph- is not on the agenda, as what will come next will
supersede this, completely, and I do mean completely.
That is, -gph- will still be there and any programs
with -gph- will work as well as they ever did, but
you won't, I guess, ever want to write another
program calling -gph- again.
What old hands know very well, and what new hands might
want explaining, is that Stata Corp policy (I should
perhaps say Stata Corp style) is, on the whole,
not to say _anything_ in public about _what_ is coming
_when_, on various grounds. (In contrast, you
get more indiscretion from Stata Corp developers
at user group meetings, which is some incentive
to attend Boston 2003, London 2003, etc.)
The grounds include these
1. Great reluctance to make promises which might not
be manageable (and there have been promises that
had to be broken, which if anything increases Corp
embarrassment and reluctance to make any announcements).
2. Disdain for hype, psychological games with
competitors, customers, etc. If other software
firms behave differently, that is their affair.
3. At any one time there are many development
projects ongoing within Stata Corp, many of which
do not make it to the next shipped release. Getting
out good software which works is sometimes the
least important issue, to be balanced with other
considerations, including the need to provide
technical support and documentation for anything
included in official Stata. The decision of what
to include in the next release sometimes comes
very late in the development cycle.
The best example from history -- perhaps prehistory
for many users -- is the Windows port. Windows
users make up the largest segment of Stata users,
yet the port to Windows came very late in the
day, for a mixture of technical and other reasons.
There was no Windows version for some years
after most of the obvious competitors had
started to support that platform.
In fact, there was a longstanding project developing Stata
for Windows even while many DOS users were saying repeatedly
"And when are you are going to do Stata for Windows?".
What took some time was the developers working their way
to an implementation that not only worked
but was agreed, at least internally, to be well designed
and would age well without needing repeated major revision.
Eventually, the long wait was ended when the Windows
port was put on the market.
The graphics project is very big. Those
who have visited Stata Corp may have seen
glimpses. Here is my take as an anthropologist
who studies the tribe. The project had to do
several quite difficult things:
1. be totally upward compatible with
existing graphics.
2. provide a good range of canned
routines
3. allow easier and more versatile low-level
programming by Stata programmers
4. allow interactive editing of graphs.
The more you know, the more you will see that
this set of desiderata imposes _lots_ of problems,
the aim of course not being to produce
something which works, but something which
is excellent and will be the basis for the
next several years.
The total consequence of this has been very far-reaching,
including very deep changes in Stata's internal code which
will in due course have implications elsewhere
within future versions.
As said, Stata Corp has a policy of keeping
quiet, so don't expect any official expansion on this,
least of all in terms of dates or details.
Nick
[email protected]
*
* 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/
