On Tue, May 28, 2013 at 9:00 AM, Don Guinn <[email protected]> wrote:
> The unicode numbers for the box characters are
>
>    9!:6 ''
> ┌┬┐├┼┤└┴┘│─

Yes, I get this.

>    3 u: 9!:6 ''
> 16 17 18 19 20 21 22 23 24 25 26

I confirm this result also.

>    3 u: 7 u: '┌┬┐├┼┤└┴┘│─'
> 9484 9516 9488 9500 9532 9508 9492 9524 9496 9474 9472

This one is interesting. Those characters are the characters from the
original line, but
   3 u: 7 u: 9!:6 ''
16 17 18 19 20 21 22 23 24 25 26

So what happened here?

I think the issue is that the OS sees those characters and provides
them back to J as unicode characters.

>    9!:7]4 u: 9484 9516 9488 9500 9532 9508 9492 9524 9496 9474 9472
> |domain error
> | 9!:7]4 u:9484 9516 9488 9500 9532 9508 9492 9524 9496 9474 9472

A related issue is:
   '┌┬┐├┼┤└┴┘│─' -: 4 u: 9484 9516 9488 9500 9532 9508 9492 9524 9496 9474 9472
0

Or, more concisely:
   (-: 4 u: 3 u: 7 u: ]) '┌┬┐├┼┤└┴┘│─'
0

Or, more concisely:
   (-: 7 u: ]) '┌┬┐├┼┤└┴┘│─'
0

I'm not completely sure why J does not allow "wchar" for 9!:7 but that
is something that should probably eventually be changed.

>    JVERSION
> Engine: j701/2011-01-10/11:2
> Library: 8.01.011
> Qt IDE: 1.0.8
> Platform: Win 64
> Installer: j801 beta install
> InstallPath: c:/j/j64-801a

I was also using J8 (which of course uses the J7 engine).

> The third line made by copying and pasting the output from the output of
> the first line. Looks like J is doing a translation from the old box
> characters to the unicode box characters. What would happen if the unicode
> characters were displayed directly instead of using the 9!:6 '' numbers?
> And maybe it would be nice to allow unicode values for 9!:7 so one could
> use rounded corners or other unicode box characters if one wanted?

As I tried to illustrate, above, I think it's the OS performing that
translation.

And I do not know the specifics of 9!:7 - why it's currently
constrained to "ascii" characters. But I suspect it has something to
do with the specific characters being used for box drawing not being
unicode characters, and a lack of generically reliable unicode font
support for box characters. (Not that box drawing characters are
completely reliable as it is...)

A related issue is that unicode is an open-ended standard so most of
its implementation probably belongs in user space.

Another related issue is that "wchar" is "insufficiently wide" for
many unicode characters.

-- 
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to