* Re: Rational ADA development environment
1990-02-10 20:10 ` Dennis Moore
@ 1990-02-12 13:08 ` David Nettles
1990-02-12 17:50 ` info wanted on debugging concurrent tasks Clement Pellerin
1990-02-20 21:55 ` Rational ADA development environment George Mitchell
1990-02-12 14:43 ` RICK CARLE
1990-02-12 17:03 ` Jeff Clark
2 siblings, 2 replies; 11+ messages in thread
From: David Nettles @ 1990-02-12 13:08 UTC (permalink / raw)
In article <4722@rtech.rtech.com> dennism@menace.UUCP (Dennis Moore) writes:
>ssawyers@cadfx.ccad.uiowa.edu (Steve Sawyers,212 ERF,,) writes:
>
>Isn't this typical for a government project?!? 40,000 LOC and 2,500 pages
>of documentation? Are you serious? 40,000 lines of code is 667 pages (at
>60 lines per page). That's almost 4 pages of documentation per line of
>code, even if your LOC counter doesn't count blank lines or comments as lines.
>
>If ADA is such a wonderful, self-documenting, easy to code, easy to understand,
>easy to maintain language (as the government claims it is), then why are 2,500
>pages of documentation necessary?
>
>-- Dennis Moore, my own opinions, blahblahblah
Nice cheap shot.
Though not familiar with the particular project, I have worked both sides of
a DoD project (military and civilian contractor). Let me list some of the
attending documents that are required as deliverables, regardless of
language used.
System Specification
System Design Document
Software Requirements Specification
Software Development Plan
Interface Requirements Specification
Software Test Plan
Interface Design Document
Software Design Document
Software Test Descriptions
Software Test Reports
Operation and Support Documents
Version Description Document
The government is going to get their pound of documentation regardless of
how wonderful the language is.
--
David Nettles
^ permalink raw reply [flat|nested] 11+ messages in thread
* info wanted on debugging concurrent tasks
1990-02-12 13:08 ` David Nettles
@ 1990-02-12 17:50 ` Clement Pellerin
1990-02-13 17:38 ` Loren Louis Hart
1990-02-20 21:55 ` Rational ADA development environment George Mitchell
1 sibling, 1 reply; 11+ messages in thread
From: Clement Pellerin @ 1990-02-12 17:50 UTC (permalink / raw)
Debugging concurrent programs is becoming a hot research topic.
I hope to adapt Ada debugger technology to Concurrent C.
Since Ada programmers have been writing concurrent programs
for a long time, I suspect there must be some nice
debuggers out there. We might learn some lessons
because for now, debugging Concurrent C
programs has been near to impossible.
What is the state of debugger technology for Ada?
I am especially interested in "methodologies" for debugging
concurrent programs, i.e some theory as to how to go about it.
Once a methodology is found, a debugger can be written to support
it. The theory does not have to be mathematical.
So how do YOU go about debugging concurrent tasks?
Is debugging tasks as hopeless as debugging concurrent C processes?
I would like a small description of what your favorite
debugger can do together with any reference to articles
that you've seen.
thank you.
--
news <clement
Clement Pellerin, McGill University, Montreal, Canada
clement@opus.cs.mcgill.ca
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: info wanted on debugging concurrent tasks
1990-02-12 17:50 ` info wanted on debugging concurrent tasks Clement Pellerin
@ 1990-02-13 17:38 ` Loren Louis Hart
0 siblings, 0 replies; 11+ messages in thread
From: Loren Louis Hart @ 1990-02-13 17:38 UTC (permalink / raw)
Most of the Ada debuggers are very nice, mostly because they are new and it
doesn't make sense to use obsolute techniques on a new system that you are
going to try to sell.
The best approach for debugging multiple tasks is in a windowing environment.
each task that you want to debug fires up it's own window and acts like it's
own debugger. There are several systems like this, the one I am most familiar
with is the Vax VMS debugger running on a Vaxstation. That particular
debugger allows you to have one debugger per process. The Vax Debugger
is not Ada spacific. I have also had good luck using the Verdix Debugger.
I am sure I have missed some other good debuggers, but it is tough to stay
current on all of them.
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Rational ADA development environment
1990-02-12 13:08 ` David Nettles
1990-02-12 17:50 ` info wanted on debugging concurrent tasks Clement Pellerin
@ 1990-02-20 21:55 ` George Mitchell
1 sibling, 0 replies; 11+ messages in thread
From: George Mitchell @ 1990-02-20 21:55 UTC (permalink / raw)
In article <438@software.software.org> nettles@software.org (David Nettles)
writes:
` ....
`>Isn't this typical for a government project?!? 40,000 LOC and 2,500 pages
`>of documentation? .... That's almost 4 pages of documentation per line of
`>code ...
`
`Nice cheap shot.
^^^^ ?
My arithmetic shows 16 LOC per page of documentation. Is this so terrible
given the requirements of DOD-STD-2167A? What does it have to do with Ada?
--
/s/ George vmail: 703/883-6029
email: mitchell@community-chest.mitre.org [alt: gmitchel@mitre.arpa]
snail: GB Mitchell, MITRE, MS Z676, 7525 Colshire Dr, McLean, VA 22102
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Rational ADA development environment
1990-02-10 20:10 ` Dennis Moore
1990-02-12 13:08 ` David Nettles
@ 1990-02-12 14:43 ` RICK CARLE
1990-02-12 17:03 ` Jeff Clark
2 siblings, 0 replies; 11+ messages in thread
From: RICK CARLE @ 1990-02-12 14:43 UTC (permalink / raw)
In article <4722@rtech.rtech.com>, dennism@menace.rtech.COM (Dennis Moore) writes:
> Isn't this typical for a government project?!? 40,000 LOC and 2,500 pages
> of documentation? ... > If ADA is such a wonderful, self-documenting, easy
> to code, easy to understand, > easy to maintain language (as the government
> claims it is), then why are 2,500 pages of documentation necessary?
DoD-Std-2167 (& 2167A) is the true culprit here. Total ignorance of the
project in question does not inhibit me from suggesting that its
excessive documentation is caused by 4 related problems.
1) Dod-Std-2167 and its insistence on too many documents with too much
detail, all to describe a single program (a 2167 CSCI).
2) The tendency of software designers to map Ada compilation units (ie,
procedures, functions, tasks, packages) to 2167 "units" (2167A CSUs),
thus producing excessive paragraphs in the 2167/A SDDD & SDD documents.
It would be better to map Ada packages (or Library Unit Groups ala
Kaye Grau/Kathy Gilroy) to 2167 units.
3) The tendency of software designers, using Ada PDL, to over-design
practically to the point of coding. This causes every line of code to
be part of the design. This tendency has always been a problem with
software designers, but Ada PDL gives them the best tool ever for
committing their sins. One solution is simple restraint, perhaps
enforced by management. A more practical solution might simply be to
design no deeper than package specs.
4) The failure of government contracting officers and industry proposal
managers to routinely tailor 2167 to the needs of the particular
contract. Proponents of 2167 have claimed (don't expect me to provide
sources of quotes) that tailoring is essential to the successful
application of 2167, but few contracts follow that doctrine.
Rick Carle
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Rational ADA development environment
1990-02-10 20:10 ` Dennis Moore
1990-02-12 13:08 ` David Nettles
1990-02-12 14:43 ` RICK CARLE
@ 1990-02-12 17:03 ` Jeff Clark
2 siblings, 0 replies; 11+ messages in thread
From: Jeff Clark @ 1990-02-12 17:03 UTC (permalink / raw)
In article <4722@rtech.rtech.com> dennism@menace.rtech.COM (Dennis Moore) writes:
> Isn't this typical for a government project?!? 40,000 LOC and 2,500 pages
> of documentation? Are you serious? 40,000 lines of code is 667 pages (at
> 60 lines per page). That's almost 4 pages of documentation per line of
> code, even if your LOC counter doesn't count blank lines or comments as
> lines.
>
> If ADA is such a wonderful, self-documenting, easy to code, easy to
> understand, easy to maintain language (as the government claims it is),
> then why are 2,500 pages of documentation necessary?
Hmmmm... When the users of *your* software systems need to learn how to
effectively make use of your products do you usually respond "Let them read
the source code..."? Wasn't somebody famous once beheaded for an attitude
like this? :-)
Jeff Clark Honeywell Systems and Research Center Minneapolis, MN
inet: jclark@src.honeywell.com
uucp: jclark@srcsip.UUCP
DISCLAIMER: If you think I speak for my employer, you need serious help ...
^ permalink raw reply [flat|nested] 11+ messages in thread