Re: [Bug-apl] Trace & stop control

[Blake McBride]

Dear Juergen,

I couldn't agree with you more.  Of course, the wacky way it does work is very convenient.  →'' has always meant do not branch, so perhaps no solution ends up doing what you'd almost always want.  →N is contradictory, I agree.

Thanks a lot for making the change!

Blake

On Wed, Jun 11, 2014 at 8:28 AM, Juergen Sauermann <address@hidden> wrote:

Hi Blake,

I changed GNU APL to behave like IBM APL2, see SVN 320.

I actually believe that the behavior shown by IBM APL2 is not very consistent.
In my opinion continuation of a stopped function should be →'' or →⍬ and not →N.
The problem with →N is that now →N in a function behaves differently (it stops) than →N
from immediate execution (doesn't stop).

But compatibility rules.

/// Jürgen

On 06/10/2014 06:26 PM, Blake McBride wrote:

It doesn't allow continuation of a stop:

GNU APL:

      ∇test

[1] '1'

[2] '2'

[3] '3'

[4] '4'

[5] ∇

      test

1

2

3

4

      S∆test←3

      test

1

2

test[3]

      →3

test[3]

The branch to 3 should have caused the continuation of the program.

IBM APL 2:

      ∇TEST

[1] '1'

[2] '2'

[3] '3'

[4] '4'

[5] ∇

      TEST

1

2

3

4

      S∆TEST←3

      TEST

1

2

TEST[3]

      →3

3

4

      )SI

      TEST

1

2

TEST[3]

Thanks.

Blake

On Mon, Jun 9, 2014 at 12:53 PM, Juergen Sauermann <address@hidden> wrote:

Hi Blake,

thanks, fixed in SVN 316.

/// Jürgen

On 06/02/2014 02:20 AM, Blake McBride wrote:

Just offering an opinion -

Since APL trace and stop are quite useful, and are part of the standard, my opinion is that these should be top priority - second only to bug fixes.  These should come before work on enhancements or fixes to extensions.

Thanks.

Blake