I tried the ravel everything method, but that lead to APL2 disclosing a
vector into a matrix APL2 fought me touth and nail. I gave up and
wrote box/unbox.
--blake
On Sat, May 17, 2014 at 11:18 PM, Frederick H. Pitts wrote:
> Blake,
>
> Thanks for the detailed response. I unders
In implementing support for this, I realise that that I'm breaking
backwards compatibility. That is, the latest version of gnu-apl-mode will
not be able to control an older version of GNU APL.
When were you planning to release GNU APL 1.4? I think it might be needed
to make sure everything is in s
I want to make sure that I correctly understand ⎕EA and ⎕EC. Should
these be able to clean up after failed execution of a defined function?
What I'm seeing is that ⎕EA and ⎕EC perform as expected given an
expression of variables and primitive functions (including expressions
that fail to execute)
That bring be back to my previous argument that the APL designers made a
mistake when they defined 'a' to be a character instead of a single-element
array. It's inconsistent, and leads to ugly code where all
single-character-strings have to be escaped as such: (,'a'), or as you put
it: (1⍴'a') whic
Blake,
Thanks for the detailed response. I understand the issue more clearly
now.
I think I would ask the user to reshape scalar arguments to vectors
(e.g. (1⍴'k')(1⍴'v')). But then, that is as onerous as having to
remember to use box everywhere its required.
Regards,
Fred
On
I tripped on that this morning.
Temporary workaround while waiting for the patch:
gnu-apl-interactive.el, line 234: change "--emacs" to "--emacs foo"
That was enough to get it going again on my systems.
On Sun, 2014-05-18 at 11:07 +0800, Elias Mårtenson wrote:
> Yes. Known problem. The interfac
Lastly,
⍴(box 'abc'),box 'def'
┌→┐
│2│
└─┘
Blake
Oh, and I should point out this too:
⍴(box 'k'),box 'v'
┌→┐
│2│
└─┘
⍴'kv'
┌→┐
│2│
└─┘
Blake
On Sat, May 17, 2014 at 9:59 PM, Blake McBride wrote:
> Greetings,
>
> I suppose my examples were meant to show that my box and unbox function as
> I hoped, and that they functioned consi
Thanks. Your APL mode for Emacs is hard to live without!
Blake
On Sat, May 17, 2014 at 10:07 PM, Elias Mårtenson wrote:
> Yes. Known problem. The interface has changed and I will fix it today.
>
> Regards,
> Elias
> On 18 May 2014 11:05, "Blake McBride" wrote:
>
>> Greetings,
>>
>> With the
Works great. Thanks!
Blake
On Sat, May 17, 2014 at 9:17 AM, Juergen Sauermann <
juergen.sauerm...@t-online.de> wrote:
> Hi Blake,
>
> thanks, fixed in SVN 270.
>
> Please note that ⎕AT is not standard and that the attributes other
> than "locked" will not be supported by GNU APL.
>
> Note als
Yes. Known problem. The interface has changed and I will fix it today.
Regards,
Elias
On 18 May 2014 11:05, "Blake McBride" wrote:
> Greetings,
>
> With the latest version of GNU APL and Emacs APL mode, I now get:
>
> Process apl exited abnormally with code 2
>
> This worked fine until I up
Greetings,
With the latest version of GNU APL and Emacs APL mode, I now get:
Process apl exited abnormally with code 2
This worked fine until I updated to the latest GNU APL. I tried rebuilding
Emacs mode native library, but it didn't help.
Thanks.
Blake
Greetings,
I suppose my examples were meant to show that my box and unbox function as
I hoped, and that they functioned consistently with all data
configurations. I wasn't intending to point out the benefits. I suppose I
thought that might be clear from my earlier emails. In response to you,
ho
BTW, I actually have a use for redefining a pendent function.
The package manager may unload itself by ⎕ex-ing all of its names. It
does this, of course, while the main package manager function is
pendent.
If that's all I did, then the main function would be missing when I
called the ]pkg user co
Looks good. Thank you.
On Sat, 2014-05-17 at 20:10 +0200, Juergen Sauermann wrote:
> Hi David,
>
> I have changed the handling of function pending on the SI. You can now
> ⎕FX functions if the pending function name is localized. This is, I believe,
> what the standard tries to say ("locally-erasa
Here's a patch to let ]usercmd accept restatement of an existing user
command definition so long as the associated function name and mode are
identical.
]usercmd ]foo foo 1
User-defined command ]foo installed.
]usercmd ]foo foo 1
⍝ without the patch there'd be a BAD COMMAND r
Hi David,
I have changed the handling of function pending on the SI. You can now
⎕FX functions if the pending function name is localized. This is, I believe,
what the standard tries to say ("locally-erasable") but does not go as
far as IBM.
I guess in practice this will be confusing anyhow. SV
Fair enough. I'll implement this tomorrow and will let you know how it went.
Regards,
Elias
On 17 May 2014 22:51, Juergen Sauermann wrote:
> Hi Elias,
>
> SVN 271.
>
> I think emacs mode is special because it survives )LOAD and is not visible
> on APL level.
>
> If we would allow that in genera
Hi Elias,
SVN 271.
I think emacs mode is special because it survives )LOAD and is not
visible on APL level.
If we would allow that in general then troubleshooting could become
rather tricky because
you cannot easily figure the state of the interpreter.
By enforcing the use native functions
Regardless of the thinking behind the GPL, I don't see the benefit of ever
locking or "hiding" a function, ever.
In fact, I can't think of any other programming language that has a feature
like that.
Regards,
Elias
On 17 May 2014 22:17, Juergen Sauermann wrote:
> Hi Blake,
>
> thanks, fixed in
Hi Blake,
thanks, fixed in SVN 270.
Please note that ⎕AT is not standard and that the attributes other
than "locked" will not be supported by GNU APL.
Note also that ⎕CR shows an empty result when a function is locked, while
10 ⎕CR and )DUMP display it even if locked.
This is because I believe
I am OK with this, and it would certainly fill all current needs for the
Emacs mode.
I personally feel that this should be more generic (ability to load more
than one library; decoupling it from the --emacs flag). The decision is
yours though, and the Emacs mode works equally well regardless of wh
Hi Elias,
my proposal would be this:
1. we give the --emacs command line option an argument. So apl would be
started like:
*apl --emacs emacs_arg*
2. the libemacs library provides a function emacs_start() with two char
* arguments:
*void emacs_start(const char * emacs_arg, const c
23 matches
Mail list logo
