Re: PARI module

[Hoehle, Joerg-C. <Joe...@t-...>]

Hi,

>it would be nice if you could look at this more closely.
>(e.g., by installing pari on your machine and building the pari =
module)
I have 0 time for this these days.
When I'll get time (next year), I'll do these of my short-term TODO
o defpackage documentation precisions
o postgres sql module patch
o FFI MEM-READ/WRITE add-ons
o add 64bit FFI testcases

>>>when "next" is (c-ptr-null entree) I get segfault whenever I call
>> What's hefty? 1 million elements?
>maybe more.
Then (c-ptr next) is certainly bad, if you're not interested in the =
whole thing.
(c-pointer entree) looks like the best match.

However, given non corrupt data structures, creating a list of a =
million elements should be handled by the code, or we could learn =
something about some weakness somewhere (testing with huge data is =
always a good thing to do).

>charset:iso-8859-1 :INPUT-ERROR-ACTION :ignore))
>(pari::is_entry "norm")
>#S(PARI::entree :NAME "norm" :VALENCE 257 :VALUE=20
>#<FOREIGN-ADDRESS #x00D22FB0>
>   :MENU 2 :NEXT #<FOREIGN-ADDRESS #x00FAFABB> :HELP "`A=C2=BAA=C2=BA"
>   :ARGS #<FOREIGN-ADDRESS #x00FBE381>)

>(setq *foreign-encoding* (make-encoding :charset charset:ascii=20
>:INPUT-ERROR-ACTION :ignore))
>(pari::is_entry "norm")
>#S(PARI::entree :NAME "norm" :VALENCE 257 :VALUE=20
>#<FOREIGN-ADDRESS #x00D22FB0>
>   :MENU 2 :NEXT #<FOREIGN-ADDRESS #x00FAFABB> :HELP "`"
>   :ARGS #<FOREIGN-ADDRESS #x00FBE381>)

So is_entry :return-type (c-ptr entree) yields these strange help =
strings, while your hack_doc() accessor yields proper ASCII?!?
(c-lines "char* hack_doc (char* s) { entree *e =3D is_entry(s); return =
e=3D=3DNULL?NULL:e->help; }~%")
That's looks like a mystery.
BTW, what's the output of (hack_doc "norm")?

Do any of the c-pointers look like they point to stack locations =
instead of heap?

Regards,
	J=C3=B6rg H=C3=B6hle