Gnat 3.11 whinings

Gnat 3.11 whinings

[dennison]

OK. The long wait is over, and I have a great new public version of gnat to
play with. Much cudos to ACT for all the hard work and commitment. Now, let
the whining begin! :-)

Am I the only one who had to add the "c:\gnat\bin" path to my file path
manually on NT to get gnat to run?

Also, I don't like the change to the Ada emacs bindings which makes gnatmake
run automaticly when C-cC-c is pressed. This is probably more convienent for
a majority of users, but it makes things a royal pain for the rest of us. For
instance, I now have to edit a file (after finding out which file to edit) to
change any of the default compilation flags. Also, we have *multiple* Ada
compilers here. We used to pick which one by modifying that compile command
line...

I think issues like this is why the standard compile.el for emacs was written
to allow you to view and change the command line before before it gets
executed. Needless to say, I found the change from the standard behavior
quite disconcerting. Yes, I did find the new "Customize" menu, with the flag
for disabling the new behavior buried at the bottom. But considering all the
recent talk about the importance of following standards, this variance from
the standard emacs practice suprising. :-)

I think either the HTML code in some of the documentation, or the "texitohtml"
translator ACT used has bugs. For instance, the first sentence of the second
paragraph of the Ada mode documentation on my Netscape 4.5 starts out:

@item full Integrated Development Environment : @itemize @bullet @item support
of 'project files'

Well, that's it for my *first* 10 minutes w/ 3.11...

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[dennison]

In article <78kus9$4nv$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> OK. The long wait is over, and I have a great new public version of gnat to
> play with. Much cudos to ACT for all the hard work and commitment. Now, let
> the whining begin! :-)

I have another, but this one is a biggie!

The NT version of the gnat installer toasted my site-start.el. Now I expect
files such as ada-mode.el, ada-stmt.el, etc. to get changed. I made sure to
put our customizations in different files to protect against this. But
site-start.el is *mine*, not yours. I can see gnat appending stuff in there,
but it should never *replace* it.

To make matters worse, we have the site-lisp directory installed on a shared
drive with everyone's emacs configuration pointing to it. So now whenever any
developer decides to install gnat...wham! There goes our site-start again.

Fortunately, I keep my feathers numbered, for just such an emergency. %-)

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[robert_dewar]

In article <78kus9$4nv$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> OK. The long wait is over, and I have a great new public
version of gnat to
> play with. Much cudos to ACT for all the hard work and
commitment. Now, let
> the whining begin! :-)

I would suggest that discussions which are specific to
GNAT will probably be more fruitful on chat@gnat.com.
There are many knowledgable and experienced users of
GNAT there, many of whom have been using 3.11 for a
while, who are not readers of CLA, and who are more
likely to be able to help you out. Also this is CLA,
not CLG! :-)

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[Al Christians]

This is not a whine  -- just a question.  I don't see anything in the
DLL section of the documentation about  how to make DLL's from Ada
sources with Gnat, although I infer from the documentation that it can
be done.  This process has previously been described as a 'black art'.  
Could anyone, perhaps even someone who already knew how to do it before 
3.11p, tell me if there is anything in the new release that could lead 
to expeditious DLL creation by the hoi polloi, eg me? If so, what,
where?

TIA for any most gratuitous advice more apropos than 'get Object Ada'.

BTW, I would consider addition of this long-awaited feature to the
public version of Gnat as a notable accomplishment toward the goal 
discussed in the 'how to make Ada more popular' thread.  However, my
question is in no way a criticism of Gnat or ACT.  I am most pleased to 
be able to obtain a copy of this product under a no-cost license, and I 
am in no position to tell them what they should do or to offer any 
informed comment about their product feature decisions.  It will be
useful to me regardless of how DLL's are supported. I'm just trying
to plan the solution of some immediate development problems that I would
like to solve with DLL's written in Ada, and would like the benefit of
the advice of those more astute about what possibilities the new 
release offers.

Thanks again.

Al

Re: Gnat 3.11 whinings

[Markus Kuhn]

robert_dewar@my-dejanews.com writes:
|> I would suggest that discussions which are specific to
|> GNAT will probably be more fruitful on chat@gnat.com.

May be worth adding that one can subscribe the chat@gnat.com
mailing list by sending

  To: chat-request@gnat.com

a message with any subject and the message body

  subscribe chat
  end

Markus

-- 
Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
Email: mkuhn at acm.org,  WWW: <http://www.cl.cam.ac.uk/~mgk25/>

Re: Gnat 3.11 whinings

[Al Christians]

robert_dewar@my-dejanews.com wrote:
> 
> I would suggest that discussions which are specific to
> GNAT will probably be more fruitful on chat@gnat.com.
> 

You are right again, Robert.  Someone posted a more succinct version
of my question there and got what looks to be a complete and explicit
answer explaining how to build Windows DLL's with GNAT.  Hooray 
and thanks.

To continue what I was saying in reference to the 'making Ada popular'
thread (which is why I decided to post my gnat wuestion here), this
is a big step ahead for GNAT and Ada.  As Jerry van Dijk noted, the 
Ada tools market is (only!) $100 million.  Given software companies
might
aim for $1 million of revenue per employee if they are selling at 
mass-market and competitive prices with healthy marketing budgets, that
would mean that the entire Ada tools industry can only support on the 
order of 100 employees (more if expenses, wages, and profits are
lower).  
This gives a pretty good indication that it will be a long time before 
Ada has everything that everybody wants (The development team for MSVC++
was more than 100 people).   So the ability to put code into DLL's, to 
give more chance forr Ada to do the things that it does well as part of
a multi-language system looks important  to getting Ada into more common 
usage.   

Kudos.

Al

Re: Gnat 3.11 whinings

[Larry Kilgallen]

In article <78l2qq$845$1@nnrp1.dejanews.com>, dennison@telepath.com writes:

> To make matters worse, we have the site-lisp directory installed on a
> shared drive with everyone's emacs configuration pointing to it. So now
> whenever any developer decides to install gnat...wham! There goes our
> site-start again.

That sounds like a good argument for file protection (and developers not
running with privileges enabled by default).

Larry KIlgallen

Re: Gnat 3.11 whinings

[dennison]

In article <78l3en$8sb$1@nnrp1.dejanews.com>,
  robert_dewar@my-dejanews.com wrote:
> In article <78kus9$4nv$1@nnrp1.dejanews.com>,
>   dennison@telepath.com wrote:
> > OK. The long wait is over, and I have a great new public
> version of gnat to
> > play with. Much cudos to ACT for all the hard work and
> commitment. Now, let
> > the whining begin! :-)
>
> I would suggest that discussions which are specific to
> GNAT will probably be more fruitful on chat@gnat.com.

OK. I've reposted the message there (I think).

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[dennison]

In article <1999Jan26.163956.1@eisner>,
  Kilgallen@eisner.decus.org.nospam wrote:
> In article <78l2qq$845$1@nnrp1.dejanews.com>, dennison@telepath.com writes:
>
> > To make matters worse, we have the site-lisp directory installed on a
> > shared drive with everyone's emacs configuration pointing to it. So now
> > whenever any developer decides to install gnat...wham! There goes our
> > site-start again.
>
> That sounds like a good argument for file protection (and developers not
> running with privileges enabled by default).

As has been hinted at in a previous thread, we aren't really into draconian
measures around here. I may technicly be the "owner" of the site-start.el
file, but developers here are allowed (and *encouraged*) to put items of
their own creation in the site-start if they will be of general use (a rather
important "if"). It would seriously violate our team spirit, plus create more
work for me, if I were to start acting as "emperor" of the group
configuration files.

In a larger group this setup would probably be a nightmare. But we only have
about 8 developers who all sit within earshot of each other, and it seems to
work well for us. For instance, one developer created an elisp routine to
convert UML specifications into Ada package specs...

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[dewar]

In article <78na8i$204$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> As has been hinted at in a previous thread, we aren't
> really into draconian measures around here. I may
> technicly be the "owner" of the site-start.el file,
> but developers here are allowed (and *encouraged*) to put
> items of their own creation in the site-start if they
> will be of general use

Well for most people putting important files under
configuration control does not constitute being "draconian"
but in any case of course you have a backup of this file.

So yes, indeed it is an annoyance to have to go and fix
the installation, by going to the backup, and we will
indeed try to fix this in a future release. Meanwhile,
we will add some documentation to warn (the rather small
number of) people who modify the site-start file that they
will have to do an extra simple manual step for now.

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[dewar]

In article <78n93n$rn$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> OK. I've reposted the message there (I think).

Just to be absolutely clear, discussions about problems
are fine on chat@gnat.com. But bug reports ONLY come to
the attention of ACT if they are sent to report@gnat.com.
So if are not sure if something is a bug and needs
discussion, then it is fine to discuss it on CLA or
chat (I am a bit surprised that T.E.D. thinks the
site.el problem he saw is in this category).

If you think something is a bug, report it as a bug
to report@gnat.com. So far we have received no reports
from T.E.D for example of problems, so as far as the
ACT radar is concerned, no problems have been reported
by him.

Actually the site-el problem is already reported, by us,
to ourselves, with an internal tracking number. it is
something we noticed during the shake down period, but
decided was not serious enough to cause what would have
been a significant (e.g. one month) additional delay
in the release, so that one at least is reported and
will be fixed in the 3.12 release.

But I have a vague memory of T.E.D reporting other
concerns, I can't remember if they were bugs or not,
but for sure they will not be looked at if they do
not come in to report@gnat.com.

This policy may seem a bit arbitrary (draconian :-) but
it is really necessary for our report control mechanisms
to work smoothly and reliably.

We really appreciate those of you who HAVE taken the
trouble to send in useful input to report from the first
release of 3.11p. Obviously the first few days can smoke
out quite a bit. So far, nothing too serious has turned up,
but quite a few little things have been reported and
already fixed.

I know it can be a bit frustrating at times that we do
not have the resources to correspond with you about
problems you submit in this way, but please be assured we
do eventually look at all problems submitted, and if you
take the time to send properly formatted clear bug reports
to report@gnat.com, you are helping everyone improve GNAT
in the long run.

Thanks again to those who have contributed in this and
other ways to the GNAT project.

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[dennison]

In article <78otom$fbs$1@nnrp1.dejanews.com>,
  dewar@gnat.com wrote:
> If you think something is a bug, report it as a bug
> to report@gnat.com. So far we have received no reports
> from T.E.D for example of problems, so as far as the
> ACT radar is concerned, no problems have been reported
> by him.

Fair enough. But I think its also reasonable to publicly warn people who
might be considering downloading the compiler about an installer bug if it is
potentially *destructive*. I certainly would have appreciated such a post
before *I* installed it.

As you say, this problem has already been discovered by ACT and logged, so my
reporting it to ACT at this point would serve no purpose whatsoever.


T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Gnat 3.11 whinings

[robert_dewar]

In article <78q0cd$bk2$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> In article <78otom$fbs$1@nnrp1.dejanews.com>,
>   dewar@gnat.com wrote:
> > If you think something is a bug, report it as a bug
> > to report@gnat.com. So far we have received no reports
> > from T.E.D for example of problems, so as far as the
> > ACT radar is concerned, no problems have been reported
> > by him.
>
> Fair enough. But I think its also reasonable to publicly
> warn people who might be considering downloading the
> compiler about an installer bug if it is
> potentially *destructive*. I certainly would have
> appreciated such a post before *I* installed it.
>
> As you say, this problem has already been discovered by
> ACT and logged, so my reporting it to ACT at this point
> would serve no purpose whatsoever.

Actually that is for the record not true. Our tracking
system most definitely differentiates between problems
we have noticed ourselves and those that users have
noticed!

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Reporting gnat bugs (was: Gnat 3.11 whinings)

[dennison]

In article <78q6ee$hj3$1@nnrp1.dejanews.com>,
  robert_dewar@my-dejanews.com wrote:
> In article <78q0cd$bk2$1@nnrp1.dejanews.com>,
>   dennison@telepath.com wrote:
> > As you say, this problem has already been discovered by
> > ACT and logged, so my reporting it to ACT at this point
> > would serve no purpose whatsoever.
>
> Actually that is for the record not true. Our tracking
> system most definitely differentiates between problems
> we have noticed ourselves and those that users have
> noticed!

Really? Having always worked on custom-developed software, that possiblity had
not occurred to me. Is it a real difference in priority, or just another
possible sort key?

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Reporting gnat bugs (was: Gnat 3.11 whinings)

[Larry Kilgallen]

In article <78qbak$m7t$1@nnrp1.dejanews.com>, dennison@telepath.com writes:
> In article <78q6ee$hj3$1@nnrp1.dejanews.com>,
>   robert_dewar@my-dejanews.com wrote:
>> In article <78q0cd$bk2$1@nnrp1.dejanews.com>,
>>   dennison@telepath.com wrote:
>> > As you say, this problem has already been discovered by
>> > ACT and logged, so my reporting it to ACT at this point
>> > would serve no purpose whatsoever.
>>
>> Actually that is for the record not true. Our tracking
>> system most definitely differentiates between problems
>> we have noticed ourselves and those that users have
>> noticed!
> 
> Really? Having always worked on custom-developed software, that possiblity had
> not occurred to me. Is it a real difference in priority, or just another
> possible sort key?

Defects which are (potentially) user-visible are more important
than those which cannot be detected by the user.

It can be difficult to prove that a defect cannot be detected by
the user, but proving the affirmative by example is trivial, if
you have the example.

Larry Kilgallen

Re: Reporting gnat bugs (was: Gnat 3.11 whinings)

[dennison]

In article <1999Jan29.102747.1@eisner>,
  Kilgallen@eisner.decus.org.nospam wrote:
> In article <78sf18$fg8$1@nnrp1.dejanews.com>, dewar@gnat.com writes:

> Also, presuming ACT has a reasonable filing system, the next time
> someone is working on a particular module it is possible for them
> to look at outstanding reports to see if any of them naturally fit
> well with what they are working on.  That might be efficient.
>
> There is absolutely no chance of them doing that if the report is
> in your file rather than theirs.

Quite true. But the issue was reporting *known* bugs, not unknown bugs.

T.E.D.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Reporting gnat bugs (was: Gnat 3.11 whinings)

[dewar]

In article <1999Jan28.163826.1@eisner>,
  Kilgallen@eisner.decus.org.nospam wrote:

> Defects which are (potentially) user-visible are more
> important than those which cannot be detected by the
> user.
>
> It can be difficult to prove that a defect cannot be
> detected by the user, but proving the affirmative by
> example is trivial, if you have the example.

Yes, exactly.

When a user reports a bug that raises its importance. The
primary difference for us in handling reports from
supported and non-supported users is the following:

1. For supported users, we are happy, often at great length
to explain how to use GNAT and how to use Ada 95, and offer
any advice we can on how to make their projects succeed.
This is service that we cannot provide for unsupported
users.

2. For defect reports, the importance of a problem is not
affected by whether someone is supported, but the *urgency*
most certainly is. For supported customers, we immediately
work to figure out work arounds, and/or correct the defect,
and/or explain why it is not a defect as quickly as
possible. For non-supported customers, defect reports may
wait a considerable while to be processed, especially if
they do not appear to be of a critical nature (which is in
fact true for most such reports).

So the bottom line here is that reporting a bug to us is
always useful in establishing the importance of the
problem, and can only help to improve future releases for
both supported and unsupported users.

Indeed the experimentation and academic research use of the
public version of GNAT has over the years been an important
source of input. This is part of the synergy of open source
software. Unsupported users help find problems before
supported users using the software for mission critical
applications run into those problems, and in general the
larger user community means that the product matures faster
for everyone.

In retrospect, for Ada 83, one of the unfortunate things
was that the user communities for early versions of Ada 83
technology, with the possible exception of Ada/Ed, were
small, and so the technologies matured more slowly. I think
anyone looking at the history here would see that Ada 95
compilation technology has matured faster than we saw in
Ada 83. Part of this is just second-time-around, but a lot
of it is the more open distribution of Ada 95 compilers.

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Re: Reporting gnat bugs (was: Gnat 3.11 whinings)

[Larry Kilgallen]

In article <78sf18$fg8$1@nnrp1.dejanews.com>, dewar@gnat.com writes:

> So the bottom line here is that reporting a bug to us is
> always useful in establishing the importance of the
> problem, and can only help to improve future releases for
> both supported and unsupported users.

Also, presuming ACT has a reasonable filing system, the next time
someone is working on a particular module it is possible for them
to look at outstanding reports to see if any of them naturally fit
well with what they are working on.  That might be efficient.

There is absolutely no chance of them doing that if the report is
in your file rather than theirs.

Larry Kilgallen

Re: Reporting gnat bugs (was: Gnat 3.11 whinings)

[robert_dewar]

In article <1999Jan29.102747.1@eisner>,
  Kilgallen@eisner.decus.org.nospam wrote:
> Also, presuming ACT has a reasonable filing system, the
> next time someone is working on a particular module it is
> possible for them to look at outstanding reports to see
> if any of them naturally fit well with what they are
> working on.  That might be efficient.

Yes, that is exactly the way things work. Also sometimes
you feel like finding an easy thing to work on, and if you
see that a (non-supported) user has run into a problem,
it makes it more interesting to look at than if it is a
purely internally identified problem.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own