comp.lang.ada
 help / color / mirror / Atom feed
* From VERDIX to GNAT
@ 2000-08-01  0:00 stephane
  2000-08-02  0:00 ` Harry Overs
  2000-08-02  0:00 ` Robert L. Klungle
  0 siblings, 2 replies; 3+ messages in thread
From: stephane @ 2000-08-01  0:00 UTC (permalink / raw)


I have to evaluate the conversion of a large project from VERDIX/HP-UX to GNAT/Linux. My main problem seems to be how to replace VERDIX libraries by GNAT ones. I haven’t found any clue along this news group, can anybody help me?

TIA

Stephane.

------
User of http://www.foorum.com/. The best tools for usenet searching.




^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: From VERDIX to GNAT
  2000-08-01  0:00 From VERDIX to GNAT stephane
@ 2000-08-02  0:00 ` Harry Overs
  2000-08-02  0:00 ` Robert L. Klungle
  1 sibling, 0 replies; 3+ messages in thread
From: Harry Overs @ 2000-08-02  0:00 UTC (permalink / raw)


I have done this on SGI, though there was a little more help available.
SGI produced bridge packages that replicated VERDIX ones.

However, your best approach is not to go for a GNAT version first off.
Depending on what you have already, and whether you have to maintain a
running system.

Its best to find the units that are not compileable/portable. Then fix these
in your Verdix build.

That is, get the state of your HP-UX software such that it is fully
portable.
Then move this over to GNAT HP-UX.
So at this stage you should have a build the will build under Verdix or Gnat
under HP-UX.
That way you can test and verify correctness of the software.

If you can get it to compile and run under GNAT HP-UX then there should be
only minor problems to solve in the linux gnat build.

Once you've done all this, then it would be good to look at the code as a
whole and take advantages of Ada 95 stuff.



However, depending on the Verdix and other stuff you use, it may not be
possible to keep a compatible Verdix and Gnat build.


If you mail the specific packages being used and how they are being used, we
can see if there are Ada 83, or 95 ways of doing these.
I have not found anything that was not possible to do purely in Ada 95 yet.

"stephane" <stephane.reiniche@steria.fr> wrote in message
news:200081-162646-12143@foorum.com...
> I have to evaluate the conversion of a large project from VERDIX/HP-UX to
GNAT/Linux. My main problem seems to be how to replace VERDIX libraries by
GNAT ones. I haven&#8217;t found any clue along this news group, can anybody
help me?
>
> TIA
>
> Stephane.
>
> ------
> User of http://www.foorum.com/. The best tools for usenet searching.






^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: From VERDIX to GNAT
  2000-08-01  0:00 From VERDIX to GNAT stephane
  2000-08-02  0:00 ` Harry Overs
@ 2000-08-02  0:00 ` Robert L. Klungle
  1 sibling, 0 replies; 3+ messages in thread
From: Robert L. Klungle @ 2000-08-02  0:00 UTC (permalink / raw)


stephane wrote:

> I have to evaluate the conversion of a large project from VERDIX/HP-UX to GNAT/Linux. My main problem seems to be how to replace VERDIX libraries by GNAT ones. I haven&#8217;t found any clue along this news group, can anybody help me?
>
> TIA
>
> Stephane.
>
> ------
> User of http://www.foorum.com/. The best tools for usenet searching.

I agree with Harry Overs.

I ported 73_000 lines from VERDIX SGI to Linux GNAT Intel.
I went directly to the Intel. In hind-sight, should have done it on SGI GNAT
first.

The three main problem areas were:
1. Record Representation Clauses (major drudgery, there were a lot of them).
2. Variant location in Variant records (simple fix once decoded the documentation.
3. Math Library (math.lib vs gnat's generics). Created package which instantiated
the generics wanted, and included it. Some function names different (e.g. asin vs
arcsin, etc).

Total port time was ~3 weeks.

Moving to Linux actually improved the situation dramatically.
The VERDIX Ada run time continually extended the dynamic memory
pool from the kernel, even though the program returned memory to
the pool during execution (there were no memory leaks). Eventually
the system ran out of memory and would crash (512MB system).
In talking to SGI and VERDIX, they said they did it for speed of the system.
Didn't expect anyone to run a system that had to stay up for days,
and sometimes weeks without reloading the program.

cheers...bob.






^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2000-08-02  0:00 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2000-08-01  0:00 From VERDIX to GNAT stephane
2000-08-02  0:00 ` Harry Overs
2000-08-02  0:00 ` Robert L. Klungle

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox