Re: Ada to C Translator

[Dave Ptacek <drptaceknospam@collins.rockwell.com>]

Wow, I guessed I really stepped in to delicate topic here.  All of the
replies had good information and points that should really be addressed
before starting an Ada to C/C++ port.  By the way Robert, thanks for
putting on the unbiased hat and defending some of the rather opinionated
views.  I knew that I was approaching blasphemy with my post considering
it was to a bunch of Ada people, but was hoping I worded it tactfully
enough so that I wouldn't step into a religious war.  I think I
succeeded on that one although maybe the group took pity on a relative
"newbie" to the group, I certainly did not mean to offend anyone.

I will once again emphasize that finding perspective new employees to
help us with our legacy Ada test equipment application is indeed a
problem (now I'm NOT speaking for the rest of the products developed
here, nor for the company in general).  Considering our test solutions
have a lifespan of 20-25 (maybe more) years, selecting a language and
lineage of toolsets has a large impact on future costs.

We started with Ada because the companies commitment to it in the
mid-80's, and even then our group bucked the trend and chose the Alsys
Ada compiler over the companies preferred toolset.  As most of you know,
Alsys essentially left the market in the early 90's and we've been
running with an unsupported, possibly obsolete, compiler ever since.  We
did have 4-5 years of support prior to them leaving which gave us a good
indication of what could be done and shouldn't be done in Ada while
using Alsys.  We certainly couldn't have achieved the level of success
without the Alsys Ada compiler.

However, we now find ourselves in the middle of the technology evolution
that seems to change every 3-6 months depending on the particular
technology of interest.  Contrary to the belief of some people, DOS is
not dead and you can still buy licenses for DOS today.  The same cannot
be said for Win3.1, Win3.11, and soon Win95.  It would be foolish to not
acknowledge the writing on the wall and stay with DOS, and considering
the widespread usage of WinNT, it seems to be the best offering from
Microsoft at this time.  I do believe Microsoft will always provide some
"upgrade" path from NT 4.0 to 2000 to etc. as there are just too many
Microsoft apps that need will need to be "upgraded" as well.  Jumping on
that bandwagon appears to have some merit.

I personally have considered a Linux solution, but the logistics haven't
been worked yet.  We have well over 200 test stations around the world
in service centers and customer facilities, coordinating a PC swap,
getting the station running and connecting it up to the internal
network(s) available appears to be a rather large effort.  And yes, most
of the networks are using NT servers.

I would be interested in some suggestions as to what languages and
toolsets would be viable alternatives for maintaining a program 15 - 20
years into the future.  Please try to take the Ada hat off before you
respond, it still might be your choice, attempt to rationalize it with
reasonable statements, consider the logistical situation noted above and
lastly this is not a "deep pocket" program so funds are limited.

I'm rambling again and have probably stepped into yet another delicate
topic...

thanks,

Dave