Re: [Gcl-devel] ANSI test Windows OPEN.* failures

[Camm Maguire]

Hi Mike!  Sorry for my confusion in my earlier post!

I see exactly the same thing you report below in Linux.

cell-error-name might still just be a placeholder symbol.  See
unixport/ansi_cl.lisp. 

error system still needs lots of work, but this should not be your
problem at present, unless I misunderstand again!

Please keep us posted.

Take care,

Mike Thomas <address@hidden> writes:

> Hi Camm.
> 
> Running those tests in "open.lsp" without any other tests loaded or
> run beforehand, gives a perfect result.  Somehow the lisp system is
> being corrupted beforehand.  I'd appreciate your letting me know what
> Linux GCL prints out at the point noted below.
> 
> The first place where a condition error appears after running "make
> test-unixport" is at SYMBOL-VALUE.ERROR.5.  When rt:*catch-errors* is
> t (the usual setting), I get this message:
> 
> ....
> Test SYMBOL-VALUE.ERROR.5 failed
> Form: (LET ((SYM (GENSYM)))
>          (DECLARE (OPTIMIZE SAFETY))
>          (HANDLER-CASE (PROGN (SYMBOL-VALUE SYM) :BAD)
>              (UNBOUND-VARIABLE (C) (ASSERT (EQ (CELL-ERROR-NAME C) SYM))
>                  :GOOD)))
> Expected value: :GOOD
> Actual value: #<CONDITIONS::INTERNAL-UNDEFINED-FUNCTION.1>.
> ....
> 
> Is that the same error message you get for SYMBOL-VALUE.ERROR.5?
> 
> Setting rt:*catch-errors* to nil as Paul suggested overnight causes
> the tests to throw to the debugger with the following error at the
> same location:
> 
> .....
> Expected value: NIL
> Actual value: (*COMPILE-FILE-PATHNAME* *COMPILE-FILE-TRUENAME*).
>   CL-TYPES-THAT-ARE-CLASSES.1 CL-TYPES-THAT-ARE-CLASSES.2
>   CL-TYPES-THAT-ARE-CLASSES.3 SYMBOL-PACKAGE.ERROR.1
>   SYMBOL-PACKAGE.ERROR.2 SYMBOL-PACKAGE.ERROR.3 SYMBOL-PLIST.ERROR.1
>   SYMBOL-PLIST.ERROR.2 SYMBOL-PLIST.ERROR.3 SYMBOL-PLIST.ERROR.4
>   SYMBOL-VALUE.ERROR.1 SYMBOL-VALUE.ERROR.2 SYMBOL-VALUE.ERROR.3
>   SYMBOL-VALUE.ERROR.4
> Error in UNLESS [or a callee]: The function CELL-ERROR-NAME is undefined.
> 
> Fast links are on: do (use-fast-links nil) for debugging
> Broken at EQ.  Type :H for Help.
>   1 (Continue) Retry loading file "doit.lsp".
>   2 Return to top level.
> dbl:CL-TEST>>
> 
> .....
> 
> 
> This does not make sense to me as "cell-error-name is allegedly defined:
> 
> ....
>  >(apropos "CELL-ERROR-NAME")
> CELL-ERROR-NAME
> CONDITIONS::CELL-ERROR-NAME  Function
> ....
> 
> 
> Trying again (si::use-fast-links nil) and with just the
> SYMBOL-VALUE.ERROR.5 test defined:
> 
> ....
> CL-TEST>(regression-test:do-tests)
> 
> Doing 1 pending test of 1 tests total.
> Error in UNLESS [or a callee]: The function CELL-ERROR-NAME is undefined.
> 
> Broken at INVOKE-DEBUGGER.  Type :H for Help.
>   1 (Continue) Return to top level.
> dbl:CL-TEST>>:bt
> 
> #0   INVOKE-DEBUGGER
> {loc0=#<conditions::internal-undefined-function.1>,loc1=nil
> ,loc2=#<conditions::i...} [ihs=23]
> #1   ERROR
> {loc0=conditions::internal-undefined-function,loc1=(:function-name un
> less :name ...} [ihs=22]
> #2   CLCS-UNIVERSAL-ERROR-HANDLER
> {loc0=:undefined-function,loc1=nil,loc2=unless
> ,loc3=(""),loc4="The function ~S i...} [ihs=21]
> #3  Computing args for  EQ {} [ihs=20]
> #4   DO-ENTRY {loc0=nil,loc1=nil,loc2=nil} [ihs=10]
> #5   DO-ENTRIES* {loc0=#<synonym stream to
> *terminal-io*>,loc1=nil,loc2="~&Doing
>   ~A pending test~...} [ihs=9]
> #6   DO-ENTRIES {loc0=#<synonym stream to *terminal-io*>} [ihs=8]
> #7   DO-TESTS
> {loc0=(do-tests),loc1=t,loc2=(do-tests),loc3=(do-tests),loc4=#\)}
> [ihs=7]
> #8   EVAL {loc0=nil,loc1=nil,loc2=nil,loc3=#<compiled-function
> do-tests>} [ihs=6
> ]
> #9   TOP-LEVEL {loc0=t,loc1="GCL (GNU Common Lisp)  2.7.0 ANSI    Nov
> 24 2004 20
> :49:38
> Source L...} [ihs=5]
> #10   FUNCALL {loc0=#<compiled-function system:top-level>} [ihs=4]
> NIL
> dbl:CL-TEST>>
> 
> ....
> 
> 
> As usual any hints gratefully accepted.  This positively reeks of the
> same kind of problem I have had previously with error condition
> instability especially in Axiom.
> 
> Cheers
> 
> Mike Thomas
> 
> 
> 
> 
> _______________________________________________
> Gcl-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/gcl-devel
> 
> 
> 

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