GNAT Error Reporting

GNAT Error Reporting

[Nick Roberts]

Well, I can vouch for the sophistication of GNAT's (3.11) error reporting,
anyway. I have only been trying it out for a short time, but it evidently
does indeed produce clear, simple messages, and suggests (usually perfectly
accurately) how to fix the error. Great! I give credit where it is due :-)

For my own compiler, I shall try to provide similarly useful messages and
suggestions. In addition, I fancy trying to put in the following features:

(a) a utility to facilitate the tailoring of the text of error messages (to
different languages, different target audiences, etc.);

(b) an interactive facility which progressively suggests 'fixes' to a
problem (most likely first);

(c) the ability for the user to choose a suggested fix, and have that fix
applied to the source text automatically;

(d) maybe a way of graphically relating error parameters to the source text
(something like little red boxes and arrows).

These are just ideas. There seem (to me) to be a lot of possibilities. Is
there any literature specifically on this subject available?

-------------------------------------
Nick Roberts

'The time has come,' the Walrus said,
  'To talk of many things:
Of shoes--of ships--and sealing wax--
  Of cabbages--and kings--
And why the sea is boiling hot--
  And whether pigs have wings.'
                        Lewis Carroll
          "Through the Looking Glass"
-------------------------------------

Re: GNAT Error Reporting

[robert_dewar]

In article <7ahkn2$kab$2@plug.news.pipex.net>,
  "Nick Roberts" <Nick.Roberts@dial.pipex.com> wrote:
> These are just ideas. There seem (to me) to be a lot of
> possibilities. Is there any literature specifically on
> this subject available?

There is a large literature on error detection and
correction in compilers. A visit to your local computer
science library will easily dig up a long list of
references!

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

Re: GNAT Error Reporting

[dennison]

In article <7ahkn2$kab$2@plug.news.pipex.net>,
  "Nick Roberts" <Nick.Roberts@dial.pipex.com> wrote:

> (c) the ability for the user to choose a suggested fix, and have that fix
> applied to the source text automatically;

DEC Ada when integrated with LSE (DEC's answer to emacs) would do this for
very simple errors (missing close parens, etc.).

T.E.D.

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

Re: GNAT Error Reporting

[robert_dewar]

In article <7ahvaa$10t$1@nnrp1.dejanews.com>,
  dennison@telepath.com wrote:
> DEC Ada when integrated with LSE (DEC's answer to emacs)
> would do this for very simple errors (missing close
> parens, etc.).


Actually around GNAT 3.06, I implemented stylized error
messages for missing and extra punctuation tokens that were
intended to drive such circuitry, but it never got done.
Now that we are doing a lot of work on GNAT/EMACS
integration, this seems like a nice thing to revisit!

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