comp.lang.ada
 help / color / mirror / Atom feed
* Gnatf Compilation
@ 1996-11-27  0:00 Nigel J. Tracey
  1996-11-27  0:00 ` Nigel J. Tracey
  1996-11-28  0:00 ` Robert Dewar
  0 siblings, 2 replies; 4+ messages in thread
From: Nigel J. Tracey @ 1996-11-27  0:00 UTC (permalink / raw)



I am trying to compile a version of gnatf which I can then
modify to use as the front-end anaylsis tool for an
automated test-case data generator for Ada programs,
but I am having a few problems. The source I'm using
is for gnat 3.07 and I'm compiling with gnat 3.07,
gcc 2.7.2 (also tried gcc 2.7.2.1) on linux 2.0.26.

I can compile gnatf fine when I combine it with the
gcc source and follow the instructions on how to build
gnat. I have copied all the various files which are
needed by the gnatfdrv.adb from the gnat distribution
except the oncs which are part of the ada library (e.g.
ada.ads a-charac.ads, etc as these are identical
to those in by /usr/adainclude) to a new directory.

I then compiled this with gnatmake -c gnatdrv.adb
which works fine compiling all the various source files.
I then run the binder and linker adding the few c object
files which are needed.

This all works find and produces the executable
gnatfdrv. Which if I run it outputs the same usage
information as gnatf. However running my version
of gnatfdrv on a ada source file causes it to
raise a constraint and output a message indicating that
a bug has been found (which I know it hasn't because
the real gnatf works fine!)

Using gdb I have found that the error is a constraint
error raised in sem_ch8.new_scope on line 3026.
Can anybody give any help as to why this problem is
occuring or suggest another way in which I might obtain
a abstract syntax tree of the ada source (with all the
semantic information) which I can then walk around
as part of my testing tool.

Any information would be a really big help, I've tried
all I can think of to sort out this problem.
Thanks very much for any help or advice.

Nigel

--------------------
Nigel J. Tracey MEng, Research Associate (High Integrity Systems Group).
Department of Computer Science,   Office: X/D016
University of York,               Tel   : [0|+44]1904 432769
York, England.                    E-Mail: njt@minster.york.ac.uk
YO1 5DD.                          URL   : http://sv1pc161.cs.york.ac.uk/~njt




^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: Gnatf Compilation
  1996-11-27  0:00 Gnatf Compilation Nigel J. Tracey
@ 1996-11-27  0:00 ` Nigel J. Tracey
  1996-11-28  0:00   ` Robert Dewar
  1996-11-28  0:00 ` Robert Dewar
  1 sibling, 1 reply; 4+ messages in thread
From: Nigel J. Tracey @ 1996-11-27  0:00 UTC (permalink / raw)



Nigel J. Tracey (njt@minster.york.ac.uk) wrote:
: I am trying to compile a version of gnatf which I can then
: modify to use as the front-end anaylsis tool for an
: automated test-case data generator for Ada programs,
: but I am having a few problems. The source I'm using
: is for gnat 3.07 and I'm compiling with gnat 3.07,
: gcc 2.7.2 (also tried gcc 2.7.2.1) on linux 2.0.26.
: <DELETED...>
: Using gdb I have found that the error is a constraint
: error raised in sem_ch8.new_scope on line 3026.
: Can anybody give any help as to why this problem is
: occuring or suggest another way in which I might obtain
: a abstract syntax tree of the ada source (with all the
: semantic information) which I can then walk around
: as part of my testing tool.
: 
: Nigel

I seem to have sorted the above problem by compiling
with the flags -gnatpg and -gnata. It still seems
to have a problem though. In the version I have compiled
if it run it with -dt -df to display the full syntax tree
it will get so far and then bomb out with a fatal error:
disk full in one of the Write_XXX commands output.adb
which in turn calls write from gnat.os_lib. I believe that
it is the write from gnat.os_lib which causes the problem.

The point at which this occurs seems to vary and it doesn't
seem to happen as often from a console as when in an xterm.
I think the write function is in libc (is this correct?)
- if so I'm using libc.so.5.4.7 if that helps any.

Without the -dt and -df flags it seems to work okay. Also
the version compiled as part of the complete gnat and
the binary version both seem to function perfectly (which
makes me think that it not a bug in by libc, but again
I'm not sure!)
Again any help would be appreciated.

Thanks,

	Nigel

--------------------
Nigel J. Tracey MEng, Research Associate (High Integrity Systems Group).
Department of Computer Science,   Office: X/D016
University of York,               Tel   : [0|+44]1904 432769
York, England.                    E-Mail: njt@minster.york.ac.uk
YO1 5DD.                          URL   : http://sv1pc161.cs.york.ac.uk/~njt




^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: Gnatf Compilation
  1996-11-27  0:00 Gnatf Compilation Nigel J. Tracey
  1996-11-27  0:00 ` Nigel J. Tracey
@ 1996-11-28  0:00 ` Robert Dewar
  1 sibling, 0 replies; 4+ messages in thread
From: Robert Dewar @ 1996-11-28  0:00 UTC (permalink / raw)



Nigel says

"Any information would be a really big help, I've tried
all I can think of to sort out this problem.
Thanks very much for any help or advice."

Except sending a question to report@gnat.com! Which is where the gnatfolks
are who could help you with your problem (which is probably trivial to
solve).





^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: Gnatf Compilation
  1996-11-27  0:00 ` Nigel J. Tracey
@ 1996-11-28  0:00   ` Robert Dewar
  0 siblings, 0 replies; 4+ messages in thread
From: Robert Dewar @ 1996-11-28  0:00 UTC (permalink / raw)



Nigel, perhaps another good place for your detailed questions on how
to build gnatf from sources, somthing that is perhaps a bit specialized
for CLA, is to subscribe to the chat@gnat.com mailing list, where such
tecnical discussions are probably more fruitful. This is a completely
open mailing list, for discussions of technical issues on GNAT, and in
fact discussions about how to build from sources have often occurred
there.

To join chat@gnat.com, send email to chat-request@gnat.com with
the text subscribe xxx, where xxx is your email adress.

This is an unofficial mailing list maintained by Ada Core Technologies,
unofficial in the sense that ACT does not participate in any formal manner.
However a few of the GNAT folks are on the list, and occasionally 
participate. 

Please note that the chat list is NOT the place to submit bug reports 
(neither is comp.lang.ada) The only way to submit bug reports on GNAT
is to send email to report@gnat.com following the format described
in gnatinfo.txt.





^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~1996-11-28  0:00 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1996-11-27  0:00 Gnatf Compilation Nigel J. Tracey
1996-11-27  0:00 ` Nigel J. Tracey
1996-11-28  0:00   ` Robert Dewar
1996-11-28  0:00 ` Robert Dewar

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox