Re: [Axiom-developer] A curious algebra failure

[Gabriel Dos Reis]

On Sat, 11 Aug 2007, Stephen Wilson wrote:

| From: Stephen Wilson <address@hidden>
| Date: 11 Aug 2007 19:52:56 -0400
| In-Reply-To: <address@hidden>
| Message-ID: <address@hidden>
| Lines: 26
| User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.4
| MIME-Version: 1.0
| Content-Type: text/plain; charset=us-ascii
| --text follows this line--
| Gabriel Dos Reis <address@hidden> writes:
| >        pmatchWithSl(s,p,al) ==
| >          s=$EmptyMode => nil
| >          s=p => al
| >          v:= ASSOC(p,al) => s=rest v or al
| >          MEMQ(p,$PatternVariableList) => [[p,:s],:al]
| >          null atom p and null atom s and_
| >                  (al':= pmatchWithSl(first s,first p,al)) and
| >            pmatchWithSl(rest s,rest p,al')
| 
| This function does not appear consistent.
| 
|    v:= ASSOC(p,al) => s=rest v or al

Good spot -- that escapted me.  

Let me try your suggestion and report later -- due to my set of safety to 3,
the build takes some time... 

| 
| This line can have the function return true instead of an alist,
| potentially spoiling the recursive calls and the subsequent binding to
| al'.  Possibly (not totally sure) it should read:
|         
|    v:= ASSOC(p,al) and s=rest v => al
| 
| Simply redefining pname and )co MONAD.spad as you suggested does not
| reveal the algebra failure you reported.  Is this occurring only upon
| a complete rebuild for you?

Yes, on complete build -- I would commit only if complete build is OK.

-- Gaby