🚀 go-pugleaf

eldenbrg@stortek.com (Dave Eldenburg) wrote:

>So, IBM says I should take all my fairly sophisticated command line
>applications and convert then to PM.  Now because of the above mentioned
>problems (i.e. RISC not having a concept of PC screen buffer or a mouse)
>the OS/2 that ends up on another platform does not support PM.  I'm
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Do you actually believe this or are you just being contrary?  The PC
screen buffer discussed here is the text mode buffer that everyone
since msdog days has directly written into. 

>looking at another major conversion just to get my code to run on the
>very same operating system. And the conversion will probably be back
>to the command line, at least till IBM get some other user interface
>working.  And IBM says this is _better_???

Pure B.S. - the programming model for OS/2 & PM is higher-level than
what you've done before.  The API will be the same on other platforms
(if they ever want to do that - expect 100 million OS/2 capable PCs
(386/486/P5/etc) to be sold in the next 5 years and only about 1
million workstations - why should IBM bother with 1% of the market
unless it buys them something?)

>I guess the days of programming for the least common denominator are
>gone.  

Absolutely - I won't miss e/vi/edlin one bit!

>      It also sounds like the days of the small time programmer are
>numbered too, at least in the IBM environment.  I don't have the

Work smarter, not harder.  You're never going to be able to compete
with companies who don't slog the same footpath as you.  If you
reinvent the wheel for each app, you're leaving out that extra feature
that might get you another N buyers.

Users are getting more sophisticated in their desires for interfaces -
buy tools that support this and let you work on the good stuff - the
reason people buy your work.

>resources to support a major conversion with every hardware platform
>change.  And can you imagine the maintenance problems!?!?

So buy a OS-platform independent interface and write your code to that.
Or go client-server (check out Notes).

Get away from the gate-level programming - work at a higher level and
optimize that which needs it.  

BTW, this isn't meant to be a flame - please take it as constructive
criticism.  Backwards compatibility is good, but we have a *really*
long way to go before computers are really usable.

Pete

Not an official statement from Honeywell - this is my opinion only.