Re: Web Development Using Ada?

["G.B." <rm-dash-bau-haus@dash.futureapps.de>]

On 31.07.13 13:44, Yannick Duchêne (Hibou57) wrote:
> Le Wed, 31 Jul 2013 13:30:23 +0200, G.B.
> <rm-dash-bau-haus@dash.futureapps.de> a écrit:
>> the difficulty of either effort will be
>> influenced by the prior project's use of standard techniques as
>> opposed to standard "technologies"
>
> Please, can you tell more on this?

In part, Michael Erdmann explains.

Moreover, the web market is prone to errors of adopting
"new technologies" early, falling prey to the persuasive
rhetoric of marketing: "It's easy!", "Latest technology", etc.
The market is considered a race, and some think that the
first one always wins.

Thus, "the market" spreads venture capital among the "new
technology things" per year and per start-up, in order to
see which one survives. That's very different from a standards
based solution approach. To produce true standard *solutions*,
one will---at least in an ideal world---start from experience,
think first, then try, then iterate. That's the same with
standard techniques: there is a lot to learn, like how to
organize active and passive components of a system, how
to layer the parts of a system, how to prepare for enlarging
the system's capacities, how to make the system adaptable,
how to integrate the system with accounting, etc. There is
not hot, now framework for that I know of, is there?

Working with standards takes time. Before that, it takes time
learning something comprehensive as a standard. Rewards come
only later. Pseudo-solutions ("technologies") will be
rewarding immediately, as they offer something very simple
for some very simple problems, and you can show something!
Guess how many simple problems are in need of solutions,
I mean technically simple problems, not simple use cases
such as online micro-payment (still largely unsolved).
How many of these are viable, economically?

Standards tend to be less simplistic, therefore less appealing
than the latest, easy to operate "technological achievement":
think of a web based framework that generates views of database
tables. Oh, it ignores, for now, the relational part of RDBMSs,
but, bla bla bla marketing bla bla bla(*).

Since maintenance-friendly use of well understood techniques
takes time (for learning and doing), produce cannot become
part of a documented software system in no time, to
be pushed out the door into the market AQAP.

But! Once you have a system that was programmed drawing on tech
knowledge collected during the last decades, you will see
managers smile when they witness a system not failing in the
event of data loss, or a system being adapted to some new use
case in one and a half day, using standard programming
techniques! (**) That's when a team had been using standard
techniques as opposed to standard (***) "technologies".


__
(*) Somewhat related: Hibernate is becoming an obstacle here
and there, as do hash based databases of data that turn out to
be relational in the end. Going back to standard relational
software, and to more boring, standard, old technique such
as JDBC. Oops.

(**) Standard programming techniques would include "patterns",
a preference for has-a, independent modules, planning for
parallel/concurrent execution, defensive I/O, etc.

(***) *standard* n. what everyone is currently in favor of,
a must.