Yes, but do you think Ken could have read the C code and verify that that's
what he intended? :-)
More power to you.
On Sat, Mar 9, 2019 at 5:28 PM Henry Rich wrote:
> I think you would be pleased to see the current parser implementation.
> The parsing table is transposed, so that each part of
I think you would be pleased to see the current parser implementation.
The parsing table is transposed, so that each part of speech indicates
which lines of the parse table it matches. This fits into 8 bits, so
the information for the 4 columns of the parse table fit into a single
32-bit word
It is unusual to make assignment an array with an internal datatype. But
it was all driven by the needs of the parser (see Dictionary, Section
II.E), and the J interpreter is built around the parser.
See file t.c, and search for ASGN, CASGN, and CGASGN. (The latter two in
the value parts, instea
I never knew that. I was relying on the ID byte.
I now use upper flag bits of the type to distinguish ASGN as
private/public, and to-name/to-noun.
I needed to distinguish named private assignments in explicit
definitions, because I preallocate symbols for them in a symbol table
that I also
ASGN is an internal type; in the value part (the ravel) of a scalar with
that type, there is a 0 or 1, indicating whether is local assignment (=.)
or global assignment (=:).
On Sat, Mar 9, 2019 at 2:52 PM Henry Rich wrote:
> Aaaghh, you caught me. That was one of the first changes I made when I
Aaaghh, you caught me. That was one of the first changes I made when I
was learning the J Engine. I knew it would print wrong, but I didn't
know how to make it right, and I hoped no one would ever notice.
I guess I have to go back and fix it, but I understand the system better
now, so maybe
Right. So I see I've not used J for far too long.
It's indeed only ever used directly, i.e. not in derived verbs.
Thanks Henri!
Op za 9 mrt. 2019 om 17:34 schreef Henry Rich :
> Reassigning N affects later execution of mod but not verbs created
> previously by mod:
>
> mod =: 1 : 'N&|@:x'
>
jkt@set1:~$ /usr/local/lib/j64-801/bin/jconsole
JVERSION
Engine: j701/2011-01-10/11:25
Library: 8.01.028
Platform: Linux 64
Installer: j801 install
InstallPath: /usr/local/lib/j64-801
+a=.'huh'
|domain error
| +a=.'huh'
exit 0
jkt@set1:~$ ja
JVERSION
Engine: j807/j64nonavx/linux
Hello everyone,
I think I have found a bug in 'convert/pjson'.
load 'convert/pjson'
dec_pjson_ '"\\f\\b"'
\ \
I believe double backslashes should be escaped first, as in:
dec_pjson_ '"\\f\\b"'
\f\b
I fixed the issue on my end by moving the double backslash to the first
substitution:
Reassigning N affects later execution of mod but not verbs created
previously by mod:
mod =: 1 : 'N&|@:x'
N =: 5
mod5double =: +: mod
mod5double 4
3
N =: 7
mod5double 4
3
mod7double =: +: mod
mod5double 4
3
mod7double 4
1
Henry Rich
On 3/9/2019 10:32 AM, J
Hi!
There's a fairly substantial problem with chapter 32 of FSOJ, as it uses
this definition for mod:
mod=: adverb : 'n&|@x'
Aside the fact that using it as defined yields a domain error (easily
solved using N instead of n), there is a problem where afterwards the
article seems to assume that r
11 matches
Mail list logo
