Re: What MFC Objects Can't be created on the Stack?
"Joseph M. Newcomer" <newcomer@flounder.com> wrote in message
news:rnlid25camd5vn5l15vfsjkru3vg0ockdp@4ax.com...
I hadn't used C++ prior to learning MFC in 1995, because all of the
existing C++ products
of the era had massive problems (for example, I think it was Zortech C++
that didn't
properly handle local scope destructors when exceptions were thrown; I
forget what bugs
Borland C++ had, but there were enough that I couldn't risk trying to
build apps that were
products on top of fundamentally unreliable substrates. By the time I got
around to
adopting C++, nearly all of its fundamental failure modes had been fixed
(even though it
was a subset of the language, the subset worked well within its
limitations).
C++ was stable before 1995. Borland C++ 3.0 (circa 1992,1993) was a stellar
product. It featured a syntax highlighting editor and precompiled headers.
Prior versions had hard-coded compiler table limited the size and complexity
of source code modules.
Trying to compete with Microsoft was the killer; they wanted a spreadsheet
(which never
worked terribly well, actually it had a PostScript driver written by a
real amateur so I
could never generate spreadsheet graphs that could be inserted in anything
else),
Actually Quattro Pro was a great spreadsheet that was way better than Lotus
1-2-3. Excel wasn't even a worthy competitor yet. Quattro Pro was the
first widely used software that used the right mouse button for context
menus.
they had
a decent document production system (Sprint) which they refused to
support, and things
started going downhill after that.
I thought their doc management system was called ObjectVision? Their word
processor efforts never did get off the ground. They did have a worthy
competitor to Works, in Reflex (flat file database), Paradox (pretty good
database), and Sidekick (PIM). But without a good word processor to anchor
the office suite, they were sunk.
Promoted Pascal when nobody wanted Pascal (well, they
called it Delphi. Not a bad system, but customer acceptance was limited)
Too ambitious,
not enough resources, pretty much went into a death spiral from which they
never
recovered.
Actually, Delphi was a successful project. It was Borland's RAD answer to
VB. It was better than VB. Not many C++ programmers were interested in
Pascal, but then again they were not interested in VB either. Anders
Heijlsberg subsequently went to Microsoft and did C#. I still wonder why
C++ programmers are interested in C# any more than they were interested in
Delphi; C# and Delphi are very similar in concept. Now that Stanley Lippman
has rewritten the Managed C++ Extensions, C++ seems to be the way "real"
programmers will program .NET. Unfortunately, C# has about 5 years head
start.
I remember when the Borland booth at trade shows was as big as
Microsoft's. In the last
one I saw them appear at, they had a 10x10 booth with a folding card
table.
Yesterday Borland announced www.turboexplorer.com for free versions of their
IDE's. It looks like once the IDE's are sold to another company, that
company will be very active. Maybe they can bring the magic back a second
time (the first time was their $49 Turbo Pascal 1.0), but it may be too
little too late. Eclipse has squeezed the IDE market.
-- David