Re: [Bug-apl] Calling Value::print() with different ⎕PW

[Elias Mårtenson]

I tried to specify a different PW like the below, but the with is still limited to something close to 80:

        const PrintContext pctx( PST_NONE, Workspace::get_PP(), 1000 );

        Value_P cr_formatted = Quad_CR::do_CR( cr_level, *value, pctx );

        out << *cr_formatted;

What did I do wrong?

Regards,

Elias

On 21 May 2014 21:50, Elias Mårtenson <address@hidden> wrote:

Cool. Thanks!

And what if I want to output without any quad-CR style at all? (I.e. the default output style ]BOXING none). Is there a cr-style I can use for that? Is it 0?

Regards,

Elias

On 21 May 2014 21:43, Juergen Sauermann <address@hidden> wrote:

Hi Elias,

done in SVN 281. I had to change your TraceData.cc due to the extra argument.
I used PrintContext (style + ⎕PP + ⎕PW) instead of print width alone so that other
output aspects can be controlled as well.

/// Jürgen

On 05/21/2014 02:54 PM, Elias Mårtenson wrote:

Yes, of course. :-) I was focused on the other solution to set a custom PW. Obviously adding pw as an argument to do_CR is a better idea.

Regards,
Elias

On 21 May 2014 20:52, "Juergen Sauermann" <address@hidden> wrote:

H Elias,

that sounds more like making the print width a parameter of do_CR()
because the left arg of ⎕CR is already the first argument of do_CR() ?

/// Jürgen


On 05/21/2014 02:33 PM, Elias Mårtenson wrote:

To clarify, What I have is a Value, and all I want to do is to get a string containing the printed form of that Value for a given quad-CR-left-hand-value. I don't want the resulting string to be wrapped (since I'm handling that on the Emacs side).

Regards,

Elias

On 21 May 2014 20:29, Elias Mårtenson <address@hidden> wrote:

I mean the number to the left of quad-cr. :-)

Regards,
Elias

On 21 May 2014 20:28, "Juergen Sauermann" <address@hidden> wrote:

Hi Elias,

not sure what you mean by 'CR-level' ?

/// Jürgen


On 05/21/2014 06:32 AM, Elias Mårtenson wrote:

Hello Jürgen,

I finally got around to attempting to implement this.

What I'm actually doing is to ensure that the output in a trace buffer (that displays the content of a variable in real-time) is not wrapped. This is, of course, because Emacs allows you to navigate around a larger buffer.

Currently, I'm simply calling the function do_CR which accepts a number being the CR-level that I want to use for display. When using your suggestion above, I effectively need to reimplement do_CR, which is actually somewhat complex.

Do you think it would be possible to alter the definition of do_CR so that it accepts the CR-level as an argument so that I don't have to copy&paste all that code into the Emacs module?

Regards,

Elias

On 18 February 2014 18:36, Elias Mårtenson <address@hidden> wrote:

Thank you. This is a lot better.

I was doing that just because I didn't see a better way to do it. Clearly, this is a better way to do it. :-)

Regards,

Elias

On 18 February 2014 18:34, Juergen Sauermann <address@hidden> wrote:

Hi Elias,

normally you do something like this:

// const Value & value;

PrintContext pctx(style, Workspace::get_PP(), Workspace::get_CT(), Workspace::get_PW());
PrintBuffer pb(value, pctx);
UCS_string ucs(pb, value.get_rank(), pctx.get_PW());
...

You can use your own ⎕PW value instead of Workspace::get_PW() above.

It is generally a bad idea if programs or functions other then the interpreter itself modify the data structures
of the interpreter because the functions of the interpreter make certain assumptions regarding the state of
its data structures. If other functions modify these data structures, then it becomes impossible for me to
maintain the interpreter.

/// Jürgen

On 02/18/2014 04:58 AM, Elias Mårtenson wrote:

In my native code, I would like to be able to get a printable representation of a Value_P as if ⎕PW was set to some very large value (effectively unlimited).

What is the most efficient way to do this?

I was trying to call assign on the return value from Workspace::get_v_Quad_PW(), but I got a DOMAIN ERROR (I suppose I made some silly mistake somewhere though) but I don't think that might be the best way to it.

Regards,

Elias