where is the faq

where is the faq

[Rick Sutcliffe]

Where is the faq for this group?

or, if someone does not mind answering a question that may be there

Where can I get a MacOS Ada 95 compiler? (Checked a lot of sites and
nada.)

Failing that, what is the best Win NT compiler?

Rick

-- 
Rick Sutcliffe
Trinity Western University (Not speaking officially)

Re: where is the faq

[David Botton]

I do not know about MacOS, but for NT there are many compilers.

Take a look at http://www.adapower.com/links.html for links to Compiler
Vendors and No cost versions of compilers.

David Botton


Rick Sutcliffe wrote in message <140719991226209634%rsutc@twu.ca>...
>Where is the faq for this group?
>
>or, if someone does not mind answering a question that may be there
>
>Where can I get a MacOS Ada 95 compiler? (Checked a lot of sites and
>nada.)
>
>Failing that, what is the best Win NT compiler?
>
>Rick
>
>--
>Rick Sutcliffe
>Trinity Western University (Not speaking officially)

Re: where is the faq

[Ted Dennison]

In article <140719991226209634%rsutc@twu.ca>,
  rsutc@twu.ca wrote:
> Where is the faq for this group?

http://www.adahome.com/FAQ/comp-lang-ada.html

But its a bit out of date. It hasn't been updated since Xmas 97.

--
T.E.D.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: where is the faq

[Ted Dennison]

In article <140719991226209634%rsutc@twu.ca>,
  rsutc@twu.ca wrote:

> Failing that, what is the best Win NT compiler?

That depends on what your critera are.

Gnat is good if you want to compile large programs and want to spend as
little as possible. Its also a good choice if you want to use other
tools in the gnu toolchain, or if you are used to a Unix command-line
interface and prefer working that way. Gnat is the *only* choice if you
want annex E support.

The Aonix compiler is a good choice if you don't need to compile large
programs, or you don't mind paying for the compiler. (If you need
support, it may even be *cheaper* than Gnat w/ support) Its also a good
choice if you want to link with MS Visual C++ object files, or if you
prefer a Visual-style IDE. It also has a win32 GUI-builder, if I recall
correctly.

--
T.E.D.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: where is the faq

[tmoran]

>> Failing that, what is the best Win NT compiler?
>
>That depends on what your critera are.
  Note also that Aonix (among others) gives you a complete walk-back
in ada.exceptions.exception_information and also in case of an
unhandled exception.  With Gnat you'll get "Constraint Error" or such
as your only clue about where the problem might lie.  So, unless you
write perfect code every time, the first time, you'll *always* need
to run Gnat exe's under the debugger.  If you want similar failure
information if your program dies on a customer, you'll have to tell
him to run your program only under the debugger.

Re: where is the faq

[Stanley R. Allen]

tmoran@bix.com wrote:
> 
>  With Gnat you'll get "Constraint Error" or such
> as your only clue about where the problem might lie.  So, unless you
> write perfect code every time, the first time, you'll *always* need
> to run Gnat exe's under the debugger.  If you want similar failure
> information if your program dies on a customer, you'll have to tell
> him to run your program only under the debugger.

I thought there was a way to get a trace with the GNAT
utility addr2line.  Is this correct?  I see the executable
for addr2line in the gnat binary directory, but there's
nothing in the standard documentation about it (unless it
is separately supplied).


-- 
Stanley Allen
mailto:s_allen@hso.link.com

Re: where is the faq

[David Kristola]

In article rsutc@twu.ca, Rick Sutcliffe <rsutc@twu.ca> () writes:
>Where is the faq for this group?

http://www.adahome.com/FAQ/comp-lang-ada.html#title

>or, if someone does not mind answering a question that may be there
>
>Where can I get a MacOS Ada 95 compiler? (Checked a lot of sites and
>nada.)

Check out Tenon's website at      http://www.tenon.com/products/
Both MachTen and CodeBuilder provide platforms for running GNAT
(which comes with the products) in a Unix environment inside a
well behaved Mac application.


>Failing that, what is the best Win NT compiler?

I have no idea, i use MachTen/GNAT on my Mac at home, and Apex and
GNAT on my Sun at work.  I like GNAT, but i am sure Aonix makes a
fine compiler too.


--djk, keeper of arcane lore & trivial fluff
Home: David95037 at aol dot com
Spam: goto.hades@welovespam.com

GNAT -vs- ObjectAda (was Re: where is the faq)

[Steve Doiel]

tmoran@bix.com wrote in message ...
>>> Failing that, what is the best Win NT compiler?
>>
>>That depends on what your critera are.
>  Note also that Aonix (among others) gives you a complete walk-back
>in ada.exceptions.exception_information and also in case of an
>unhandled exception.  With Gnat you'll get "Constraint Error" or such
>as your only clue about where the problem might lie.  So, unless you
>write perfect code every time, the first time, you'll *always* need
>to run Gnat exe's under the debugger.  If you want similar failure
>information if your program dies on a customer, you'll have to tell
>him to run your program only under the debugger.

This is exactly why we're using ObjectAda right now over GNAT.  In our
code we quietly write the exception information to a file and keep the
system running.  We can then correct the cause of exceptions at our
leisure.  I believe that GNAT would require that the program be stopped
in the debugger.

BTW: I have heard that in some time in the future (don't know when)
similar traceback information will be available for GNAT.

We have compiled and ran our code under both ObjectAda and GNAT.
The code compiled under GNAT tends to run approximately 2X the
speed.  The GNAT debugger (GDB) also permits viewing nearly all
variables, while the ObjectAda debugger is somewhat limited in this
regard, which is frustrating.

At some time in the not to distant future (I hope) Aonix will be releasing
a new version of ObjectAda (7.2) that is supposed to fix most of their
debugger woes.  This release is already more than 6 months late.

SteveD

Re: where is the faq

[Robert Dewar]

In article <DJJj3.291$Yp6.13826@typhoon-sf.snfc21.pbi.net>,
  tmoran@bix.com wrote:
> >> Failing that, what is the best Win NT compiler?
> >
> >That depends on what your critera are.
>   Note also that Aonix (among others) gives you a complete
walk-back
> in ada.exceptions.exception_information and also in case of an
> unhandled exception.  With Gnat you'll get "Constraint Error"
or such
> as your only clue about where the problem might lie.  So,
unless you
> write perfect code every time, the first time, you'll *always*
need
> to run Gnat exe's under the debugger.  If you want similar
failure
> information if your program dies on a customer, you'll have to
tell
> him to run your program only under the debugger.

This is out of date information. The latest version of GNAT
for NT can give full traceback information if you use zero
cost exception handling.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: where is the faq

[Robert Dewar]

In article <DJJj3.291$Yp6.13826@typhoon-sf.snfc21.pbi.net>,
  tmoran@bix.com wrote:
> >> Failing that, what is the best Win NT compiler?
> >
> >That depends on what your critera are.
>   Note also that Aonix (among others) gives you a complete
walk-back
> in ada.exceptions.exception_information and also in case of an
> unhandled exception.  With Gnat you'll get "Constraint Error"
or such
> as your only clue about where the problem might lie.  So,
unless you
> write perfect code every time, the first time, you'll *always*
need
> to run Gnat exe's under the debugger.  If you want similar
failure
> information if your program dies on a customer, you'll have to
tell
> him to run your program only under the debugger.

This is out of date information. The latest version of GNAT
for NT can give full traceback information if you use zero
cost exception handling.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Robert Dewar]

In article <3790b774.0@news.pacifier.com>,
  "Steve Doiel" <nospam_steved@pacifier.com> wrote:
In our
> code we quietly write the exception information to a file and
keep the
> system running.  We can then correct the cause of exceptions
at our
> leisure.  I believe that GNAT would require that the program
be stopped
> in the debugger.

In fact it is quite easy to write a debugger script that would
achieve this automatically, several of our customers have used
such scripts. There is no need to stop the program.

> BTW: I have heard that in some time in the future (don't know
when)
> similar traceback information will be available for GNAT.

Various traceback capability is indeed available in GNAT version
3.12. I will post an announcement about GNAT 3.l2 today or
tomorrow.

Robert Dewar
Ada Core Technologies


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Ted Dennison]

In article <3790b774.0@news.pacifier.com>,
  "Steve Doiel" <nospam_steved@pacifier.com> wrote:
> We have compiled and ran our code under both ObjectAda and GNAT.
> The code compiled under GNAT tends to run approximately 2X the
> speed.  The GNAT debugger (GDB) also permits viewing nearly all
> variables, while the ObjectAda debugger is somewhat limited in this
> regard, which is frustrating.

Ahh, yes. One important thing I left out of the comparison list.
ObjectAda doesn't have many options for optimization. Apparently about
all it can do is some peephole optimizations. gcc (and thus gnat) has
*tons* of optimization options. I wouldn't be too suprised to hear that
this often causes optmized Gnat exe's to run faster than optimized OE
exe's.

--
T.E.D.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Gautier]

> We have compiled and ran our code under both ObjectAda and GNAT.
> The code compiled under GNAT tends to run approximately 2X the
> speed.

Same rate experienced when choosing a new compiler for a commercial
program (~100'000 LOC). What is surprising is that the old Alsys
compiler seemed to produce compact and fast code... maybe not
fit for present bloatware demand...

-- 
Gautier

--------
http://members.xoom.com/gdemont/

Re: where is the faq

[tmoran]

> The latest version of GNAT for NT can give full traceback
> information if you use zero cost exception handling.
  Excellent!  I look forward to using it!

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Pascal F. Martin]

In article <7mqtl0$jvr$1@nnrp1.deja.com>,
	Robert Dewar <robert_dewar@my-deja.com> writes:
> In fact it is quite easy to write a debugger script that would
> achieve this automatically, several of our customers have used
> such scripts. There is no need to stop the program.

This still assumes you run under debugger, which is highly annoying,
especially when the user is not a programmer. Robert, just try helping
a non computer savvy guy run through your procedure, and this _over
the phone_ !

BTW, is GDB able to detach from a process under NT ? VC++ v5 cannot.
If not, that means you have to keep the program under debug once you
looked at it. Very annoying if the program is to be run permanently
(or whatever approximation of "permanent" exists on NT..).

-- 

Pascal F. Martin.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Robert Dewar]

In article <HzUm3.481$U4.31394@clnws01.we.mediaone.net>,
  pascmartin@mail.earthlink.net (Pascal F. Martin) wrote:
> This still assumes you run under debugger, which is highly
> annoying, especially when the user is not a programmer.
> Robert, just try helping a non computer savvy guy run through
> your procedure, and this _over the phone_ !

Well obviously if a non-programmer is to run the program under
GDB in this manner, the proper approach is to have a programmer
construct an appropriate script so this is automatic, something
that is quite easy to do, and which we have helped a number of
people to do.

> BTW, is GDB able to detach from a process under NT ? VC++ v5
> cannot. If not, that means you have to keep the program under
> debug once you looked at it. Very annoying if the program is
> to be run permanently (or whatever approximation of
> "permanent" exists on NT..).

Not clear why this is annoying (though anything can annoy
anyone of course :-) GDB is entirely non-intrusive, so as long
as the program does not hit a breakpoint or get a fatal signal,
the fact that gdb is somehwhere is virtual memory has little
effect on the operating environment.

Of course it is helpful to be able to get symbolic tracebacks
without the debugger present, and this capability now exists
in many versions of GNAT (unfortunately not yet in the NT
version, which is as usual hampered by the lack of documentation
from Microsoft, but we expect to have this feature in the next
version of NT). Also in the NT version, you can indeed get
a non-symbolic traceback (using the zero cost exception support)
which can be easily converted to symbolic form using addr2line.

Robert Dewar
Ada Core Technologies



Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Pascal F. Martin]

In article <7ni275$kfs$1@nnrp1.deja.com>,
	Robert Dewar <dewar@gnat.com> writes:
> Not clear why this is annoying (though anything can annoy
> anyone of course :-) GDB is entirely non-intrusive, so as long
> as the program does not hit a breakpoint or get a fatal signal,
> the fact that gdb is somehwhere is virtual memory has little
> effect on the operating environment.

Having gdb running is annoying because:

Gdb and the application symbols takes a significant amount of memory.
That is memory the application cannot use anymore.

A PC application must be easy to install. When software is installed
on thousands of PCs, every installation problem may become costly.
If gdb has to be installed with each application, there is one more
chance something would goes wrong, especially when the user is
installing several programs, each trying to install gdb, possibly
different versions of gdb.

It is my experience that there is not two identical PC. May sound
strange, but the same model may encounter different problems. Just
the order in which the same applications are installed may (and
often do) end in different configurations.

If gdb is running, there is possibly one more task in the task bar
(for the gdb interface). One uninformed user could by mistake kill
gdb. Let remember that most people now use PCs, but computer
programmers are a minority: most PC users have no computer degree.

-- 

Pascal F. Martin.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Ted Dennison]

In article <3IRn3.743$U4.54316@clnws01.we.mediaone.net>,
  pascmartin@mail.earthlink.net (Pascal F. Martin) wrote:

> gdb. Let remember that most people now use PCs, but computer
> programmers are a minority: most PC users have no computer degree.

Heck, most *programmmers* have no computer degree.

--
T.E.D.


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Pascal F. Martin]

In article <7nsg90$iqi$1@nnrp1.deja.com>,
	Ted Dennison <dennison@telepath.com> writes:
> In article <3IRn3.743$U4.54316@clnws01.we.mediaone.net>,
>   pascmartin@mail.earthlink.net (Pascal F. Martin) wrote:
> 
>> gdb. Let remember that most people now use PCs, but computer
>> programmers are a minority: most PC users have no computer degree.
> 
> Heck, most *programmmers* have no computer degree.

Let me rephrase that:

most PC users have no computer exposure whatsoever. They usually
learned how to use Microsoft Word and hardly more. Sometime not
even that.

And that's OK: computers would not have enjoyed their current success
if high school education was required for their use.

My company, which is providing centralized train control system (no
Ada: all C, C++ and Tcl), sometime must start user training with a
dumbed down course on "what is a mouse, how to move it and what is this
strange arrow on the TV set". These persons then will control complete
mass transit systems from their multihead PCs.

No kidding.

-- 

Pascal F. Martin.

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[jerry]

Pascal F. Martin <pascmartin@mail.earthlink.net> wrote:

: My company, which is providing centralized train control system (no
: Ada: all C, C++ and Tcl), sometime must start user training with a
: dumbed down course on "what is a mouse, how to move it and what is this
: strange arrow on the TV set". These persons then will control complete
: mass transit systems from their multihead PCs.

: No kidding.

Which region did you say this was... ?

<g>

-- 
-- Jerry van Dijk | Leiden, Holland
-- Team Ada       | jdijk@acm.org
-- see http://stad.dsl.nl/~jvandyk

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Samuel T. Harris]

Robert Dewar wrote:
> 
> In article <_Fxp3.2745$U4.115180@clnws01.we.mediaone.net>,
>   pascmartin@mail.earthlink.net (Pascal F. Martin) wrote:
> > My company, which is providing centralized train control
> > system (no Ada: all C, C++ and Tcl), sometime must start user
> > training with a dumbed down course on "what is a mouse, how to
> > move it and what is this strange arrow on the TV set". These
> > persons then will control complete mass transit systems from
> > their multihead PCs.
> 
> No, there is never a situation in which you "must" follow this
> very unwise procedure which is designed to produce disasters.
> The idea that you can educate people to be effective programmers
> in this way is a manager's fantasy. You may get people to the
> point where they can write somewhat-working code, but that's
> about it.

I don't believe his company is trying to train mass transit
controllers. I believe his company is trying to train mass transit 
controllers on how to use a new PC centric system to do what they 
already know how to do. 

Having fielded a command and control system in the past, my users 
were replacing grease-boards and mission folders with PC terminals 
and screen projectors. Some of the users have _very_ little to no 
experience with PCs so we also had to start with the "basics" of
computer usage. Since we modelled the mission folders on the 
screen after the actual paper folders and the projected greaseboards
used the same format on the wall as the originals, transition
was not difficult. We were not there to train them on how
to run an aerial port, but how to use the new system to
do what they already knew how to do.

> 
> An analogous statement would be something like "our company
> is in the business of putting on piano recitals in major
> concert halls. But most typical folks don't know know anything
> about music. Sometimes we must start training concert pianists
> with dumbed down courses that start out at the level of "what
> is the difference between a white key and a black key on the
> piano". Yes, you might be able to train people who could play
> the notes more or less correctly, but you will not get competent
> musicians this way.

This statement is not analogous at all. Training in playing
the piano is not what is being attempted. Perhaps training
an already accomplished pianist in using a different kind
of keyboard would be a somewhat closer analogy.

> 
> Proper training is an essential for quality code. Training
> involves MUCH more than just learning to code in some computer
> language. In particular, programmers need an introduction to the
> notions of formal reasoning, especially if they are going to be
> working in safety critical real time areas.

You'll get no argument from me here!
I just don't believe it applies in Pascal's situation.

> 
> Robert Dewar
> Ada Core Technologies
> 
> Sent via Deja.com http://www.deja.com/
> Share what you know. Learn what you don't.

-- 
Samuel T. Harris, Principal Engineer
Raytheon, Scientific and Technical Systems
"If you can make it, We can fake it!"

Re: GNAT -vs- ObjectAda (was Re: where is the faq)

[Robert Dewar]

In article <_Fxp3.2745$U4.115180@clnws01.we.mediaone.net>,
  pascmartin@mail.earthlink.net (Pascal F. Martin) wrote:
> My company, which is providing centralized train control
> system (no Ada: all C, C++ and Tcl), sometime must start user
> training with a dumbed down course on "what is a mouse, how to
> move it and what is this strange arrow on the TV set". These
> persons then will control complete mass transit systems from
> their multihead PCs.

No, there is never a situation in which you "must" follow this
very unwise procedure which is designed to produce disasters.
The idea that you can educate people to be effective programmers
in this way is a manager's fantasy. You may get people to the
point where they can write somewhat-working code, but that's
about it.

An analogous statement would be something like "our company
is in the business of putting on piano recitals in major
concert halls. But most typical folks don't know know anything
about music. Sometimes we must start training concert pianists
with dumbed down courses that start out at the level of "what
is the difference between a white key and a black key on the
piano". Yes, you might be able to train people who could play
the notes more or less correctly, but you will not get competent
musicians this way.

Proper training is an essential for quality code. Training
involves MUCH more than just learning to code in some computer
language. In particular, programmers need an introduction to the
notions of formal reasoning, especially if they are going to be
working in safety critical real time areas.

Robert Dewar
Ada Core Technologies


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.