[Axiom-developer] Re: Bug#346552: Axiom unsoundness

[Camm Maguire]

Greetings!  Please be aware that in most cases this message is,
unfortunately, misleading about memory corruption.  Segfaults are
grouped here with a number of other errors which the error handler
cannot properly process, usually because of some bug in the error
handler.  Of course you can see if you really have a segfault with
strace, presuming you are not running with sgc on (si::sgc-on
nil). (or of course when running under gdb).

Obviously this needs improving, but an error system overhaul must wait
for 2.7.  If you can track it down to some specific gcl read error or
some such, we might be able to get a fix in 2.6.8.  I'd suggest seeing
if such an error could be triggered with AXIOMsys first, and if
necessary, check with (si::readline-off).  But at least the parent
issue appears to primarily concern sman.

Take care,

Gabriel Dos Reis <address@hidden> writes:

> Waldek Hebisch <address@hidden> writes:
> 
> [...]
> 
> | All this is based on system trace.  AFAICS when AXIOMsys runs
> | standalone there is no select loop.  I tried to give "(1) ->\n"
> | and then ^C to standalone AXIOMsys -- the trace looks quite similar
> | but AXIOMsys produces reasonable error messages.  Still, giving a
> | few bad inputs interspesed with ^C, I was able to generate:
> | 
> |    >> System error:
> |    Caught fatal error [memory may be damaged]
> 
> I've been running into this "memory may be damaged" stuff very often
> these days ]with students, everything is possible :-)].  Most of them
> happens on invalid syntax and such.  The system should be more
> resilient should not corrupt memory just because of syntax errors and
> such.
> 
> -- Gaby
> 
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah