Re: Support Metrics for Ada / C Projects

Re: Support Metrics for Ada / C Projects

[Richard Riehle]

"Dan Ligett" <ligett@softstarsystems.com> wrote in message
news:Vq2dnTTuS7iIZrTcRVn-uA@adelphia.com...
>
> In 1987, 1988, and 1989 Barry Boehm and Walker Royce reported on a version
> of Ada COCOMO they developed based on their experience at TRW.  The
> presentations were given at the annual COCOMO meetings -- I don't know if
> they were published anywhere else.
>
> I don't think they reported exactly the numbers you want to have, but if
you
> use their Ada COCOMO equations and plug in your SLOC counts, I think you
can
> calculate the productivity rates you're after.
>
Unfortunately, most of the published metrics are related to Ada 83,
not Ada 95.   There is a definite need for this kind of data, but
it tends to be closely held by the companies using Ada and they
are not as open about sharing it as we would like.

Re: Support Metrics for Ada / C Projects

[Sean Cleary]

"Richard  Riehle" <adaworks@earthlink.net> wrote in message news:<xIOYc.3067$w%6.2389@newsread1.news.pas.earthlink.net>...
> "Dan Ligett" <ligett@softstarsystems.com> wrote in message
> news:Vq2dnTTuS7iIZrTcRVn-uA@adelphia.com...
> >
> > In 1987, 1988, and 1989 Barry Boehm and Walker Royce reported on a version
> > of Ada COCOMO they developed based on their experience at TRW.  The
> > presentations were given at the annual COCOMO meetings -- I don't know if
> > they were published anywhere else.
> >
not metrics, just a caution:

5 or more years ago a project I was associated with failed because the
Ada part of the team was crippled: no run time debug -- all reported
back stuff was machine/assembly language level. They took to heavily
instrumenting their code, and got huge printouts of exactly what was
happening. And had to look through it line by line, loop cycle by loop
cycle until the problem was found. That was so slow that it killed the
project. Please do not let this happen to you. Their conclusion was
that C++ with good tools would have been better. I hope that there are
now good tools for embedded Ada.

Sean

Re: Support Metrics for Ada / C Projects

[Jeff C,]

"Sean Cleary" <seanearlyaug@hotmail.com> wrote in message 
news:1c5ac0c6.0408311010.1cde35ca@posting.google.com...
> "Richard  Riehle" <adaworks@earthlink.net> wrote in message 
> news:<xIOYc.3067$w%6.2389@newsread1.news.pas.earthlink.net>...
>> "Dan Ligett" <ligett@softstarsystems.com> wrote in message
>> news:Vq2dnTTuS7iIZrTcRVn-uA@adelphia.com...
>> >
>> > In 1987, 1988, and 1989 Barry Boehm and Walker Royce reported on a 
>> > version
>> > of Ada COCOMO they developed based on their experience at TRW.  The
>> > presentations were given at the annual COCOMO meetings -- I don't know 
>> > if
>> > they were published anywhere else.
>> >
> not metrics, just a caution:
>
> 5 or more years ago a project I was associated with failed because the
> Ada part of the team was crippled: no run time debug -- all reported
> back stuff was machine/assembly language level. They took to heavily
> instrumenting their code, and got huge printouts of exactly what was
> happening. And had to look through it line by line, loop cycle by loop
> cycle until the problem was found. That was so slow that it killed the
> project. Please do not let this happen to you. Their conclusion was
> that C++ with good tools would have been better. I hope that there are
> now good tools for embedded Ada.
>
> Sean

One time this guy I know used C++ and his dog exploded. Don't let this 
happen to you. I hope dogs are more resilant now.

I hate to feed the trolls but indicating that Ada tools for embedded were 
bad 5 (or more?) years ago is not very helpful. There were bad Ada tools 5 
years ago. There were bad C++ tools 5 years ago. In fact, there are almost 
certainly more bad C++ tools than Ada tools (given that there are more C++ 
tools in existance).

I have been using Ada since 1987. There were good tools then (VAX Ada was 
pretty nice. Even for the embedded DEC OS). I've seen some bad tools too 
(The Merdian Ada to C front end for the i860 comes to mind).

Re: Support Metrics for Ada / C Projects

[Simon Wright]

"Jeff C," <jcreem@yahoo.com> writes:

> "Sean Cleary" <seanearlyaug@hotmail.com> wrote in message 

> > 5 or more years ago a project I was associated with failed because the
> > Ada part of the team was crippled: no run time debug -- all reported
> > back stuff was machine/assembly language level. They took to heavily
> > instrumenting their code, and got huge printouts of exactly what was
> > happening. And had to look through it line by line, loop cycle by loop
> > cycle until the problem was found. That was so slow that it killed the
> > project. Please do not let this happen to you. Their conclusion was
> > that C++ with good tools would have been better. I hope that there are
> > now good tools for embedded Ada.

> I hate to feed the trolls but indicating that Ada tools for embedded were 
> bad 5 (or more?) years ago is not very helpful. There were bad Ada tools 5 
> years ago. There were bad C++ tools 5 years ago. In fact, there are almost 
> certainly more bad C++ tools than Ada tools (given that there are more C++ 
> tools in existance).
> 
> I have been using Ada since 1987. There were good tools then (VAX Ada was 
> pretty nice. Even for the embedded DEC OS). I've seen some bad tools too 
> (The Merdian Ada to C front end for the i860 comes to mind).

The VxWorks gdb from ACT with Ada support is pretty good. Looks very
nice with the (supported) GPS front-end, too!

I don't use it all that often, though ...

-- 
Simon Wright                               100% Ada, no bugs.