The two following inputs cause APL to crash:
$ echo "**" | apl --script
$ echo "10?" | apl --script
Any characters like !,*,~ and , can cause the first one. Starts happening after
about 18 characters.
The second one can be used by large numbers being passed to t
I just spent two hours isolating the test case for the )CONTINUE crash; I'm
done for today.
Note that the isolated test I posted to )CONTINUE thread does not tickle
the )IN crash...
On Thu, Jun 12, 2014 at 12:27 PM, David Lamkins wrote:
> Related to the report about )CONTINUE crashing when the
Here's the smallest test case I've been able to isolate:
file: foo.apl
⍝!
∇foo;cmd_info
cmd_info←⍬
cmd_info←cmd_info, ⊂('help' '[command]' '')
cmd_info←cmd_info, ⊂('ident' '' '')
cmd_info←cmd_info, ⊂('packages' '' '')
cmd_info←cmd_info, ⊂('read' 'package-name' '[document-id]')
cmd
)CLEAR
CLEAR WS
∇z←test z
[1] z
[2] ∇
Duplicate symbol z!!!
DEFN ERROR+
∇z←test z
^
∇x test x
[1] x
[2] ∇
Duplicate symbol x!!!
DEFN ERROR+
∇x test x
^
∇z←test[z] x
[1] x
[2] ∇
The last definition should have given a DEFN error
Blak
Greetings,
Reviewing the docs for axis argument function, it seems to me that this
feature is a lot more powerful than the docs would lead one to believe.
The docs imply that the axis argument is designed for, and perhaps can
only be used as, an axis argument. This is shown in the documentation.
In wslib5/HTML.apl there's a comment:
⍝ )COPY 3 HTML
... which doesn't match the current location of the library.
--
"The secret to creativity is knowing how to hide your sources."
Albert Einstein
http://soundcloud.com/davidlamkins
http://reverbnation.com/lamkins
http://reverbnation.c
Related to the report about )CONTINUE crashing when the package manager is
loaded...
When the package manager is in the workspace,
)out foo
succeeds.
A subsequent
)in foo
crashes APL.
Again, I'll try to find a root cause tonight unless you've already sussed
it by loading the pack
)HELP shows
...
)MORE [lib] ws-name
...
--
"The secret to creativity is knowing how to hide your sources."
Albert Einstein
http://soundcloud.com/davidlamkins
http://reverbnation.com/lamkins
http://reverbnation.com/lcw
http://lamkins-guitar.com/
http://lamkins.net/
http://successful-li
This is a preliminary report. I haven't isolated a root cause.
When my package manager is in the workspace, doing )CONTINUE crashes APL.
Furthermore, APL can't be used (it launches, but never becomes ready for
input) until CONTINUE.xml is removed.
I'll see what I can do tonight to come up with a
The first line of )CHECK output goes to CERR, while the other two go to the
stream passed as a function argument. Is this the intended behavior?
--
"The secret to creativity is knowing how to hide your sources."
Albert Einstein
http://soundcloud.com/davidlamkins
http://reverbnation.com/lamki
Hi Elias,
OK, I will do nothing.
BTW I have separated the )commands from the APL statements in
Command::process_line().
/// Jürgen
On 06/12/2014 05:58 PM, Elias Mårtenson wrote:
That might be a solution. However, let's not do it just yet. It may be
that once I have gotten the rest of the
That might be a solution. However, let's not do it just yet. It may be that
once I have gotten the rest of the user interface in place there may be a
more obvious thing that should be done.
Regards,
Elias
On 12 Jun 2014 17:55, "Juergen Sauermann"
wrote:
> Hi Elias,
>
> normally )CLEAR does the
Hi Elias,
normally )CLEAR does the job. Could be that some .so files remain loaded
but that should
not matter. So I could simply redirect )OFF to )CLEAR under android.
/// Jürgen
On 06/12/2014 05:48 PM, Elias Mårtenson wrote:
Well, this is arguably not the biggest issue right now. ☺ It is,
Well, this is arguably not the biggest issue right now. ☺ It is, however,
something I came across during my experimentation.
I have to admit that I'm not sure what is the logical thing to do when
receiving the )OFF command. Android applications generally never exits. I
suppose doing a full restart
Hi Elias,
we have the HAVE_ANDROID macro already, so I can return from )OFF
without doing anything under android.
Note that cmd_OFF is called from other places as well, not
sure how they react if we return instead of exit()-ing.
/// Jürgen
On 06/12/2014 05:20 PM, Elias Mårtenson wrote:
On And
On Android, one does not want )OFF to actually kill the process itself,
since this instantly kills everything, which is problematic for an Android
application.
Of course, I can check the command typed in the input field before sending
it to the interpreter, but with the arrival of commands being e
Hi David,
thanks, fixed in SVN 323.
/// Jürgen
On 06/11/2014 07:10 PM, David Lamkins wrote:
That's a nice addition. We get the benefits of access to session
commands without a bunch of new quad-foos.
I don't really care one way or the other about providing access to
usercmds. One the one h
17 matches
Mail list logo
