In article <[log in to unmask]>, Stigers, Greg [And]
<[log in to unmask]> writes
>X-no-Archive:yes
>A legacy application is one that has been in production for more than
>twenty-four hours, and cannot now be taken back out of production.
>
>Only in our industry is the term legacy pejorative. Ironically, I could not
>find "legacy" in "jargon file" http://www.tuxedo.org/~esr/jargon/.
>
>I would suggest that legacy is generally used to refer to something that
>uses a method, technique, technology, that has somehow been replaced by
>something newer, and substantively and qualitatively different, so that were
>the same system or application developed or deployed today using some notion
>of current practices (and there's the rub), it would probably be done with
>the newer item. So a legacy computer is an older model.
Legacy goes with 'inherited'. Generally meaning that somebody pivotal
concerned with the operation or use of the system, either in IT or in
line management generally, was not part of the team who originally
conceived, designed and installed it. And therefore has no 'buy-in' to
it, and is free to criticize it for its failings, real or imagined.
And indeed what system, old or new, does not have some failings
somewhere that can be fastened on?
We are seeing more and more examples in this list of people who have
found an HP3000 chugging away in the corner of a department they have
just joined, or who have been bequeathed one (that concept again) in a
reorganisation or merger.
On asking 'what have I got here, exactly?' they are told they have a
reliable, low-maintenance application server that they can, largely,
talk to in UNIX-speak, and 'if it ain't broke, don't fix it'. And if it
is broke, it is probably very easy to fix.
And they breathe a sigh of relief, and go away, but you kind of know
that the apps will be migrating to a farm of NT servers in the next two
years. Kind of 'better the devil you know than the angel you don't', but
what can you do?
Pete Atkin sang, of the National Steel and of the old bluesmen it
evokes, 'there are dead men still alive in that guitar'.
Another aspect of legacy systems is that there are often 'departed
(though not generally actually dead) men still alive in that system'.
We have at least three cases (no names, no pack drill) of companies
using our manufacturing system where downsizing and internal changes
mean that the base of expertise that we tapped in order to customize
those systems to their needs is long gone, or out of the loop for some
other reason.
But the basic needs have not changed, and so the systems run on, like an
airliner left on autopilot where the crew have long since bailed out.
And a passenger, going up front and discovering this, touches anything
at all at his peril.... :-)
> A legacy application
>is "procedural" instead of "object-oriented" or "component-based", and may
>use a language that does (or whose practitioners do not) not easily support
>some newer methods, or which for instance uses an older screen technology
>that only allows users to proceed in certain ways (although this is great
>for heads-down data entry) or does not offer interface features most now
>take for granted (although no one has yet explained to my satisfaction the
>difference between a drop down list or radio button drawn with characters on
>a 80 x 24 grid versus a drop down list or radio button drawn in pixels).
>
I suppose I have to say <plug> 'ScreenJet' here. How about we split the
difference, and have drop down lists and radio buttons drawn in pixels,
but on an 80 x 24 grid? Perfect for laying over an existing VPLUS app,
especially if the technology driving the drop down boxes is also
overlaid.
That way the underlying app still thinks the user got the data the
'legacy' way, by relying on feats of memory, or lists taped to the wall,
or pre-prepared input sheets, and is happy.
All of which is to say that it is probably not the *way* the features
are presented, but the likelihood of whether they will be there. How
*do* you take an existing VPLUS app and add 'drop box' functionality,
which sits rather ill with its 'fill in the whole screen and press
ENTER' mode (which as you say, *is* terrific for heads-down)?
<end plug>
>In some cases, it is perfectly legitimate to refer to the previous iteration
>of the technology as legacy, but usually without the pejorative sense. The
>extreme weirdness to me is when technologies that were and still to some
>degree are "hot" "become" legacy. Is C or C++ legacy? Is ODBC legacy? Is
>client-server legacy? Sun and perhaps even Microsoft might have us believe
>so. We have Java, and OLE and JDBC, and internet front ends now.
>
They are legacy as soon as someone wants to use newer stuff, or is
disinclined to support the older stuff...
>There have been no silver bullets.
>Just as important, most of these "new" technologies are built on a
>foundation of tried and true techniques successfully used in the previous
>generation. Java was based on C++. OO is the next evolution of structured
>programming.
> It seems as if Remus and Romulus would now slay the wolf that
>raised and nurtured them.
Those who will not learn from history are condemned to repeat it. Oses
come and go, as do languages, but algorithms stay the same. The 'oldest'
books in my collection are my Knuths, which are as true now as they ever
were, but it is amazing how users of the newer tools throw the baby of
good programming practices out of the window with the bathwater of
obsoleted approaches to system design....
> Network servers are looking more like
>minicomputers, and minicomputers are looking more like mainframes, and
>mainframes continue their own slower evolutions. It is an arrogance that
>practitioners of some new technology think that they are Minerva, that they
>sprang fully formed from Zeus's head.
And Citrix/WTS is making PCs act like dumb terminals once again, and ASP
is bureau/timeshare by another name, as is FM, and Superdome is bureau
too, in its charging model, but you have to manage the damn thing for
them....
Soon (I predict) we will see the personal computer reinvented again,
under a whizzy-dizzy new name. On some platform or other.... :-)
--
Roy Brown 'Have nothing on your systems that you do not know to be
Affirm Ltd useful, or believe to be beautiful'. (After) Wm Morris.
|
