some things recently noticed

[Sherm Ostrowsky]

1.  Minor matter.  In the MathPiper IDE, if you open more than 10 windows, all windows after the ninth receive the number 0 and replace the previous one.  Is that a bug, or intentional?

2. Middling matter.  In MathPiper docs, I pick just about any scripted function, and scroll down to where it says "view source code".  When I hit that item, I get no source code.  Instead I am invariably presented with the definition of the Factors() function (!).  From anywhere.

3. Major matter.  I recall that several years ago I used to have a sporadically recurring problem whose symptoms were: after working in MathPiper for a few hours, things would begin to fail to work, randomly-seeming.  When I checked, I always found that one or more standard functions that were working earlier had stopped working.  And the never again worked correctly until I nuked the installation, downloaded a fresh checkout and rebuilt from scratch.

Well, it happened this evening.  The function that suddenly stopped working was DivisorsList().  But that is not where the real problem was, as witness the error message:

In> DivisorsList(120)
Result: Exception
Exception: In function: DestructiveInsert.
bad argument number 2(counting from 1) :

The offending argument ***( Length(1)+1 )*** evaluated to ***( Length+1 )***
 In function: DestructiveAppend, 

But this time, the problem persisted after I completely checked out a new system and built it from scratch in a fresh directory.   
It is still there.

Now what?  My system is non useable.

Sherm

[grzesiek]

Hi,

On Saturday, October 22, 2011 8:58:45 AM UTC+2, shermo wrote:

Instead I am invariably presented with the definition of the Factors() function (!).  From anywhere.

Karma :)

 

Grzesiek

[Ted Kosan]

Sherm wrote:

> 1.  Minor matter.  In the MathPiper IDE, if you open more than 10 windows,
> all windows after the ninth receive the number 0 and replace the previous
> one.  Is that a bug, or intentional?

Hmmm. I just tried opening numerous text areas/buffers in my copy of
MathPiperIDE and I was able to open 20+ of them with no problems. What
file extensions were you giving to these buffers?

> 2. Middling matter.  In MathPiper docs, I pick just about any scripted
> function, and scroll down to where it says "view source code".  When I hit
> that item, I get no source code.  Instead I am invariably presented with the
> definition of the Factors() function (!).  From anywhere.

This is a bug that I introduced into the experimental version of the
system we are working with a few weeks ago and I will fix it when I
get time.

> 3. Major matter.  I recall that several years ago I used to have a
> sporadically recurring problem whose symptoms were: after working in
> MathPiper for a few hours, things would begin to fail to work,
> randomly-seeming.  When I checked, I always found that one or more standard
> functions that were working earlier had stopped working.  And the never
> again worked correctly until I nuked the installation, downloaded a fresh
> checkout and rebuilt from scratch.
>
> Well, it happened this evening.  The function that suddenly stopped working
> was DivisorsList().  But that is not where the real problem was, as witness
> the error message:
>
> In> DivisorsList(120)
> Result: Exception
> Exception: In function: DestructiveInsert.
> bad argument number 2(counting from 1) :
>
> The offending argument ***( Length(1)+1 )*** evaluated to ***( Length+1 )***
>  In function: DestructiveAppend,
>
> But this time, the problem persisted after I completely checked out a new
> system and built it from scratch in a fresh directory.
> It is still there.
>
> Now what?  My system is non useable.

I just did an update and an ant *all and I cannot reproduce this error:

-------------------
MathPiper version js.045.
Enter an expression after any In> prompt and press <enter> or
<shift><enter> to evaluate it.

In> DivisorsList(120)
Result: {1,2,4,8,3,6,12,24,5,10,20,40,15,30,60,120}
-------------------

Did you evaluate any expressions before calling DivisorsList() or did
you call DivisorsList() immediately after launching MathPiperIDE?

Ted

[Sherm Ostrowsky]

On Sat, Oct 22, 2011 at 10:11 AM, Ted Kosan <ted....@gmail.com> wrote:

Sherm wrote:

> 1.  Minor matter.  In the MathPiper IDE, if you open more than 10 windows,
> all windows after the ninth receive the number 0 and replace the previous
> one.  Is that a bug, or intentional?

Hmmm. I just tried opening numerous text areas/buffers in my copy of
MathPiperIDE and I was able to open 20+ of them with no problems. What
file extensions were you giving to these buffers?

I just opened a lot of .mpw files from scripts4.
 

Yes.  Apparently, something I did in my program caused this.  It worked before running the program, and failed after.  Now I will try to see what happened.  The program (so far) SEEMS to be too simple to cause this mess, but it obviously did!

Sherm

[Sherm Ostrowsky]

Well, after a lot of testing, I have come to a tentative conclusion as to what caused this.  The evidence is all indirect, but pretty convincing to me.  And if my surmise is correct, it is of interest to this group because it relates to discussions we have been having.

Reducing the offending program to its bare bones, but still keeping the "crash" behavior, I found that I could apparently (pending more extensive test runs) turn off the errors by renaming some variables.  Specifically, I had LOCAL declared variables named "factor", "factors", and "lc".  Because they were declared Local, I had thought that they would not interfere with mathpiper functions of the same names.  But apparently they did.  After I changed the names, I have not seen the bad behavior -- so far.

Was I just stupid, or is this something that should not occur?

Sherm

[Ted Kosan]

Sherm wrote:

> Well, after a lot of testing, I have come to a tentative conclusion as to
> what caused this.  The evidence is all indirect, but pretty convincing to
> me.  And if my surmise is correct, it is of interest to this group because
> it relates to discussions we have been having.
>
> Reducing the offending program to its bare bones, but still keeping the
> "crash" behavior, I found that I could apparently (pending more extensive
> test runs) turn off the errors by renaming some variables.  Specifically, I
> had LOCAL declared variables named "factor", "factors", and "lc".  Because
> they were declared Local, I had thought that they would not interfere with
> mathpiper functions of the same names.  But apparently they did.  After I
> changed the names, I have not seen the bad behavior -- so far.
>
> Was I just stupid, or is this something that should not occur?

In theory, this should not occur. Could you please send me the code
that caused the crashing so I can analyze its behavior using the
Netbeans debugger?

Thanks,

Ted